Transmitter and Receiver
Architectures for Molecular
Communications: A Survey
on Physical Design With
Modulation, Coding, and
Detection Techniques
This article focuses on the design of transmitters and receivers for molecular
communication (MC). It also reviews existing literature on transmitter and receiver
architectures for realizing MC, including both nanomaterial-based nanomachines
and/or biological entities.
By MURAT KUSCU , Student Member IEEE, ERGIN DINC , Member IEEE,

BILGESU A. BILGIN , Member IEEE, HAMIDEH RAMEZANI , Student Member IEEE,
AND OZGUR B. AKAN , Fellow IEEE

ABSTRACT | Inspired by nature, molecular communications

(MC), i.e., the use of molecules to encode, transmit, and

receive information, stands as the most promising communica-

tion paradigm to realize the nanonetworks. Even though there

has been extensive theoretical research toward nanoscale

MC, there are no examples of implemented nanoscale MC

networks. The main reason for this lies in the peculiarities

of nanoscale physics, challenges in nanoscale fabrication,

and highly stochastic nature of the biochemical domain of

envisioned nanonetwork applications. This mandates develop-

Manuscript received November 5, 2018; revised March 13, 2019; accepted
April 30, 2019. Date of publication May 30, 2019; date of current version July 19,
2019. This work was supported in part by the European Research Council (ERC)
Projects MINERVA under Grant ERC-2013-CoG #616922 and MINERGRACE under
Grant ERC-2017-PoC #780645. (Corresponding author: Murat Kuscu.)

M. Kuscu, E. Dinc, B. A. Bilgin, and H. Ramezani are with the Electrical
Engineering Division, Internet of Everything (IoE) Group, Department of
Engineering, University of Cambridge, Cambridge CB3 0FA, U.K. (e-mail:
mk959@cam.ac.uk; ed502@cam.ac.uk; bab46@cam.ac.uk; hr404@cam.ac.uk).

O. B. Akan is with the Electrical Engineering Division, Internet of Everything
(IoE) Group, Department of Engineering, University of Cambridge, Cambridge
CB3 0FA, U.K., and also with the Next-generation and Wireless Communications
Laboratory (NWCL), Department of Electrical and Electronics Engineering, Koc
University, 34450 Istanbul, Turkey (e-mail: oba21@cam.ac.uk).

Digital Object Identifier 10.1109/JPROC.2019.2916081

ing novel device architectures and communication methods

compatible with MC constraints. To that end, various transmit-

ter and receiver designs for MC have been proposed in the

literature together with numerable modulation, coding, and

detection techniques. However, these works fall into domains

of a very wide spectrum of disciplines, including, but not

limited to, information and communication theory, quantum

physics, materials science, nanofabrication, physiology, and

synthetic biology. Therefore, we believe it is imperative for the

progress of the field that an organized exposition of cumulative

knowledge on the subject matter can be compiled. Thus, to fill

this gap, in this comprehensive survey, we review the existing

literature on transmitter and receiver architectures toward

realizing MC among nanomaterial-based nanomachines and/or

biological entities and provide a complete overview of mod-

ulation, coding, and detection techniques employed for MC.

Moreover, we identify the most significant shortcomings and

challenges in all these research areas and propose potential

solutions to overcome some of them.

KEYWORDS | Coding; detection; Internet of Bio-Nano

Things (IoBNT); modulation; molecular communications (MCs);

nanonetworks; receiver; transmitter.

1302 PROCEEDINGS OF THE IEEE | Vol. 107, No. 7, July 2019

0018-9219 © 2019 IEEE. Translations and content mining are permitted for academic research only. Personal use is also permitted,
but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.

https://orcid.org/0000-0002-8463-6027
https://orcid.org/0000-0001-6982-206X
https://orcid.org/0000-0002-6282-4027
https://orcid.org/0000-0003-3813-5077
https://orcid.org/0000-0003-2523-3858


Kuscu et al.: Transmitter and Receiver Architectures for MCs

I. I N T R O D U C T I O N

Molecular communications (MC) is a bioinspired com-
munication method that uses molecules for encoding,
transmitting, and receiving information, in the same way
by which the living cells communicate [1]. MC is inher-
ently biocompatible, energy-efficient, and robust in physi-
ological conditions. Therefore, it has emerged as the most
promising method to realize nanonetworks and Internet of
Bio-Nano Things (IoBNT), which defines the artificial net-
works of nanoscale functional units, such as nanobiosen-
sors and engineered bacteria, integrated with the Internet
infrastructure [1], [2]. In that respect, MC is promising
for novel applications, especially toward information and
communication technology (ICT)-based early diagnosis
and treatment of diseases, such as continuous health mon-
itoring, smart drug delivery, artificial organs, and lab-on-a-
chip [3], [4] [see Fig. 1(a)]. It bears a significant potential
as an alternative to conventional wireless communications,
especially in those environments where the latter may fail,
such as intrabody medium [5] and confined channels such
as pipe networks [6], [7]. However, the discrete nature of
information-carrying agents (i.e., molecules), peculiarities
arising from the nanophysical and biochemical processes,
and computational and energy-based limitations of com-
municating nanomachines give rise to novel challenges.
This necessitates rethinking conventional ICT tools and
devising new ones for MC in light of envisioned IoBNT
applications.

MC has been extensively studied from various aspects
over the last decade. The research efforts are mainly cen-
tered around developing information theoretical models
of MC channels [8], devising modulation and detection
techniques [9], [10], and system theoretical modeling of
MC applications [11]. For the physical design of MC trans-
mitter (MC-Tx) and MC receiver (MC-Rx), mainly, two
approaches have been envisioned: biological architectures
based on engineered bacteria enabled by synthetic biology
and nanomaterial-based architectures that are conceptu-
ally visualized in Fig. 1(b) [1]. However, none of these
approaches could be realized yet, and thus, there is no
implementation of any artificial microscale/nanoscale MC
system to date. As a result, the overall MC literature mostly
relies on assumptions isolating the MC channel from the
physical processes regarding the transceiving operations,
leading to a plethora of ICT techniques, feasibility, and
performance of which could not be validated.

Our objective in this paper is to help close the gap
between theory and practice in the MC research by pro-
viding a comprehensive account of the recent proposals
for the physical design of MC-Tx/Rx and the state-of-
the-art theoretical studies covering modulation, coding,
and detection techniques. We provide an overview of the
opportunities and challenges regarding the implementa-
tion of MC-Tx/Rx and corresponding ICT techniques that
are to be built on these devices. Throughout this review,
we concentrated mostly on the diffusion-based MC, where

the transmitted molecules propagate through passive dif-
fusion along concentration gradients. This is the most
widely utilized MC configuration in the literature, as the
propagation of molecules does not necessitate additional
complexity or energy consumption. Moreover, diffusion is
the main molecular transport mechanism in many of the
widespread natural MC systems, such as synaptic commu-
nication, quorum sensing, and Ca2+ signaling. However,
throughout this review, we also partly cover other MC
configurations, such as diffusion-based MC with the addi-
tional flow, microfluidic MC, bacteria conjugation-based
MC, and molecular-motor powered MC, while discussing
the physical design approaches for MC-Tx/Rx.

We first investigate the fundamental requirements for
the physical design of microscale/nanoscale MC-Tx and
MC-Rx, such as those regarding the energy and molecule
consumption, computational complexity, and operating
conditions. In light of these requirements, we cover the
two design approaches, namely, the nanomaterial-based
approach enabled by the newly discovered nanomateri-
als, e.g., graphene, and the biological approach enabled
by the synthetic biology tools. For nanomaterial-based
MC-Tx, we investigate architectures based on microflu-
idics, stimuli-responsive hydrogels, and nanoporous struc-
tures, whereas for biological MC-Tx, we particularly focus
on transmission schemes based on bacterial-conjugation,
virus transmission, genetic circuit-regulated protein trans-
mission, and enzyme-regulated Ca2+ transmission. For
nanomaterial-based MC-Rx, although we mostly focus on
the nanoscale field-effect-transistor biosensor (bioFET)-
based designs, we also discuss receiver architectures that
are widely utilized in initial macroscale MC experiments.
Toward the biological design of MC-Rx, we provide a brief
review of synthetic biology tools that are available for
sampling and decoding molecular messages.

In this paper, we also provide an overview of the
modulation, coding, and detection methods proposed for
MC. Modulation techniques in MC fundamentally differ
from that in conventional electromagnetic (EM) commu-
nications, as the modulated entities, i.e., molecules, are
discrete in nature and the developed techniques should
be robust against highly time-varying characteristics of
the MC channel, as well as the inherently slow nature of
the propagation mechanisms. We cover MC modulation
techniques that encode information into the concentration,
type, ratio, release time, and release order of molecules,
as well as the base sequences of nucleotides. We also
review the MC channel coding techniques that over-
come the extremely noisy and intersymbol interference
(ISI)-susceptible nature of MC channels via introducing
the redundant bits. Our review covers MC-specific chan-
nel coding methods, such as the ISI-free coding scheme
employing distinguishable molecule types, as well as the
classical coding schemes, such as block and convolution
codes adapted to MC. Finally, we review the state-of-
the-art MC detection techniques. Detection is, by far,
the most studied aspect of MC in the literature. However, in

Vol. 107, No. 7, July 2019 | PROCEEDINGS OF THE IEEE 1303



Kuscu et al.: Transmitter and Receiver Architectures for MCs

Fig. 1. (a) Intrabody continuous healthcare application of IoBNT enabled by MC nanonetworks [1]. (b) Components of an MC system with

biological- and nanomaterial-based MC-Tx/Rx design approaches, which are reviewed in Sections II and III.

devising detection methods, the lack of any MC-Rx imple-
mentation has led the researchers to make simplifying
assumptions about the sampling process, receiver geome-
try, channel, and reception noise. Therefore, we investigate
MC detection methods under two categories: detection
with passive/absorbing receivers and detection with reac-
tive receivers. We provide a qualitative comparison of
these methods in terms of considered channel and receiver
characteristics, complexity, type of required channel state
information (CSI), and performance.

In summary, this paper provides: 1) comprehensive
design guidelines for the physical implementation of
microscale/nanoscale MC-Tx/Rxs using available nanoma-
terials or biological tools; 2) a comparative review of
the state-of-the-art MC modulation, coding, and detection
techniques with an evaluation in terms of performance,
complexity, and feasibility for the envisioned MC-Tx/Rx
architectures; and 3) a detailed account of the design,
modeling, and fabrication challenges and future research
directions. We believe that this comprehensive review will
tremendously help researchers to close the long-standing
gap between theory and practice in MC, which has, so far,
severely impeded the innovation in this field linked with
huge societal and economic impacts.

The remainder of this paper is organized as follows.
In Section II, we investigate the design requirements of
an MC-Tx and review the physical design options for
nanomaterial-based and biological MC-Tx architectures.
We focus on the opportunities for the physical design of
an MC-Rx in Section III. We provide an overview of the
state-of-the-art MC modulation and coding techniques in
Section IV. A comprehensive review of the existing MC
detection schemes for passive, absorbing, and reactive
receivers is presented in Section V. In Section VI, we out-
line the challenges and future research directions toward
the implementation of MC-Tx/Rxs and the development

of the corresponding coding, modulation, and detection
techniques. Finally, we conclude this paper in Section VII.

II. M O L E C U L A R C O M M U N I C AT I O N
T R A N S M I T T E R

MC-Tx encodes information in the physical proper-
ties of molecules, such as concentration, type, ratio,
order or release time, and releases information molecules
(IMs) accordingly. To this aim, information to be transmit-
ted is required to be mapped to a sequence of bits through
source and channel coding to represent the information
with less number of bits and to introduce additional bits
to the information with the purpose of providing error
correction, respectively. Then, the modulator unit encodes
the information in the property of molecules and con-
trols the release of IMs according to a predetermined
modulation scheme. Finally, a power source and an IM
generator/container are required to provide energy and
IM molecules for MC-Tx. The interconnection of these
components in an MC-Tx is illustrated in Fig. 2. In this
section, we first discuss the requirements of an MC-Tx and
investigate the utilization of different IMs. Then, we review
the available approaches in the physical design of
MC-Txs, which can be categorized into two main groups:
1) nanomaterial-based artificial MC-Tx and 2) biological
MC-Tx based on synthetic biology.

A. Design Requirements for MC-Tx

1) Miniaturization: Many novel applications promised
by MC impose size restrictions on their enabling devices,
requiring them to be microscale/nanoscale. Despite the
avalanching progress in the nanofabrication of bioelec-
tronics devices over the last few decades [12], fabrication
of fully functional nanomachines capable of networking

1304 PROCEEDINGS OF THE IEEE | Vol. 107, No. 7, July 2019



Kuscu et al.: Transmitter and Receiver Architectures for MCs

Fig. 2. Physical architecture of an MC-Tx.

with each other and their surroundings in order to accom-
plish the desired task still evades us [13]. Added to the
technical difficulties of assembling a working machine at
nanoscale is the requirement of powering this machine
via an energy harvesting (EH) module, which is imposed
by the infeasibility of deploying a battery unit at these
dimensions. Moreover, with miniaturization, the surface
area-to-volume ratio increases, causing surface charges to
become dominant in molecular interactions. This causes
the behavior of molecules passing through nanoscale open-
ings to be significantly different than that observed in
larger dimensions [14]. Consequently, as the release aper-
tures of many considered molecular transmitter architec-
tures are commonly in nanoscale, peculiarities arising from
this phenomenon need to be taken into account in the
transmitter design.

2) Molecule Reservoir: The size restriction introduces
another very important problem for any transmitter archi-
tecture, namely, the limited resources of IMs [15]. Typ-
ically, a transmitter module would contain reservoirs,
where the IMs are stored to be released. However,
at dimensions in question, without any replenishment,
these reservoirs will inevitably deplete, rendering the
nanomachine functionally useless in terms of MC.
Moreover, the replenishment rate of transmitter reser-
voirs has a direct effect on achievable communication
rates [16], [17]. Possible theoretically proposed scenar-
ios for reservoir replenishment include local synthesis of
IMs, e.g., use of genetically engineered bacteria whose
genes are regulated to produce the desired transmit-
ter proteins [18], as well as employment of transmitter
harvesting methods [19]. Corresponding technological
breakthroughs necessary for the realization of these
approaches are emerging with advancements in the rel-
evant fields of genetic engineering [20] and materials
engineering [21], [22], respectively.

3) Biocompatibility: Most profound applications of nan-
odevices with MC networking capabilities, such as neural
prosthetics, tissue engineering, targeted drug delivery,
BMIs, and immune system enhancement, require the
implantation of corresponding enabling devices into bio-
logical tissue [23]. Combined with the increased toxicity
of materials at nanoscale [24], the issue of biocompatibility
of these devices stands out as one of the most challenging

research issues to be addressed. Moreover, corrosion of
implants inside the body is another commonly known issue
restricting device durability [25], which has an amplified
effect at nanoscale due to the large surface-to-volume
ratio. In addition to biochemical toxicity and durability,
flexibility of implanted devices is another important aspect
of biocompatibility affecting device lifetime. The mechani-
cal mismatch between soft biological tissue and implanted
device triggers inflammation, which is followed by the
scar tissue formation [26], and, eventually, restricts the
implant to carry out its intended task. Accordingly, the last
decade has witnessed extensive research on novel bio-
compatible materials, such as polymers [25], soft organic
electronics [27], dendrimers, and hydrogels [28]. It is
imperative to note that all issues of biocompatibility men-
tioned here specifically apply to nanomaterial-based device
architectures, and they can be, almost completely, avoided
by the choice of biological architecture, e.g., genetically
engineered organisms [1]. Host immune response to such
organisms is prone to shorten their in-body lifetime as
MC nanodevices, which is an observed phenomenon in the
case of cells and tissues engineered from biomaterials [29].
On the other hand, in case of successful adaptation to
the host immune system, one major concern in utilizing
genetically engineered organisms is the possible risk of
infection of the host by the introduced organisms. This risk
may be minimized by the introduction of suicidal genes to
the organisms [30] yet the possibility of random mutations
always remains.

4) Transmission Performance: Performance of a molec-
ular transmitter should also be evaluated by how much
control it has over the transmission of molecules. Impor-
tant performance metrics include OFF-state leakage, trans-
mission rate precision, resolution, and range, as well as
transmission delay. The OFF-state leakage, or unwanted
leakage of molecules from the transmitter, from the MC
perspective degrades communication significantly by con-
tributing to background channel noise. Moreover, it is
unacceptable for some niche applications, such as tar-
geted drug delivery [31], where leakage of drug mole-
cules would cause undesired toxicity to the body. During
operation, control over the rate of transmission is essen-
tial. For instance, in neural communications’ informa-
tion is encoded, among other fashions, in the number
of neurotransmitters released at a chemical synapse, and
accordingly, transmitter architectures envisioned to stim-
ulate neurons via the release of neurotransmitters, e.g.,
glutamate [32], will need to encode information in neuro-
transmitter concentrations via very precise transmissions.
In this respect, transmission rate precision, resolution,
and range are all variables that are determinant in the
capacity of communication with the recipient neuron.
Finally, the transmission delay is, in general, a quantity
to be minimized, as it contributes to the degradation in
communication rates. Moreover, for instance, in neural
communications, where information is also encoded in

Vol. 107, No. 7, July 2019 | PROCEEDINGS OF THE IEEE 1305



Kuscu et al.: Transmitter and Receiver Architectures for MCs

the frequency of signals, there is an upper bound on the
permissible transmission delay of a transmitter in order
to satisfy a given lower bound on maximum frequency
transmittable.

B. Physical Design of MC-Tx

MC uses molecules for information transfer, unlike the
traditional communication systems that rely on EM and
acoustic waves. Therefore, the physical design of MC-Tx
significantly differs from traditional transmitters. The key
components of an MC-Tx are illustrated in Fig. 2. The
first element in MC-Tx is the information source that
represents either a bitwise information generated by a
nanomachine or a biological signal from living cells, such
as neurons and cardiomyocytes.

The next building block in an MC-Tx is the processing
unit, which performs coding, modulation, and control of
IM release. This block may not appear in biological systems
utilizing MC in a basic way, such as hormonal commu-
nication inside a human body, where the information is
transferred via a single-molecule type with on–off keying
(OOK). For this purpose, MC-Tx processing unit can oper-
ate via chemical pathways [6] or through biocompatible
microscale processors or via synthetic genetic circuits [33].
In case of biological Tx architectures, processing units
performing logic gates and memory functions in the form
of synthetic genetic circuits can be embedded into cells.

Inside the processing unit, the first step is to represent
analog or digital information with the least number of
bits possible through source coding. Then, the channel
coding block introduces extra bits to make data trans-
mission more robust against error-prone MC channels.
After the coding step, digital information is transformed
into an analog molecular signal according to a predeter-
mined modulation scheme. There are various modulation
schemes proposed for MC by using molecular concentra-
tion, type of molecules, or time of molecule release. The
detailed discussion of modulation techniques for MC can
be found in Section IV-A. The processing unit controls
IM release according to the data being transmitted and
the modulation scheme. IM molecules are provided from
IM generator/container, which can be realized as either
a reservoir containing genetically engineered bacteria to
produce IM molecules on demand or a container with a
limited supply of IM molecules.

There are two types of molecule release mechanisms:
instant release, where a certain number of molecules (in
mol) are released to the medium at the same time, and
continuous release, where the molecules are released with
a constant rate over a certain period of time (mol/s).
Instant release can be modeled as an impulse symbol,
while continuous release can be considered as a pulse
wave. According to application requirements and channel
conditions, one of the release mechanisms can be more
favorable. Bitwise communications, for example, require
rapid fade-out of IMs from the channel, as these molecules

will cause ISI for the next symbol; thus, the instant release
is more promising. In addition, alarm signals for warning
plants and other insects can favor continuous release to
increase detection probability and reliability.

The processing unit in MC-Tx may require a power
source for performing coding and modulation operations,
especially in the nanomaterial-based MC-Txs. Battery-
powered devices suffer from limited lifetime and random
battery depletions, i.e., unexpected exhaustion of sensor
battery due to hardware or software failure, that signif-
icantly affect the reliability of systems that rely on MC.
Therefore, MC-Tx unit is required to operate as a self-
sustaining batteryless device by harvesting the energy
from its surrounding. For this purpose, EH from var-
ious sources, such as solar, mechanical, and chemical,
can be utilized in MC-Tx architectures. Hybrid EH tech-
niques can be employed as well to increase the out-
put power reliability [34]. Concerning the intrabody and
body area applications, the human body stands as a
vast source of energy [35], which has been exploited to
power biomedical devices and implants in many ways, e.g.,
thermoelectric EH from body heat [36], vibrational EH
from heartbeats [37], and biochemical EH from perspira-
tion [38]. Nevertheless, design and implementation of EH
methods at microscales/nanoscales for MC still stand as
the important open research issues in the literature.

1) Information-Carrying Molecules: Reliable and robust
MC necessitates chemically stable IMs that can be selec-
tively received. In addition, environmental factors, such
as molecular deformation due to enzymes and changes
in pH, can have severe effects on IMs [39] and degrade
the performance of MC. Next, we provide a discussion on
several types of IMs that can be utilized for MC.

a) Nucleic acids: Nucleic acids, i.e., deoxyribonucleic
acid (DNA) and ribonucleic acid (RNA), are promising
candidates as IMs by being biocompatible and chemically
stable, especially in in-body applications. In nature, DNA
carries information from one generation into another. RNA
is utilized as messenger molecules for intercellular com-
munication in plants and animals to catalyze biological
reactions, such as localized protein synthesis and neuronal
growth [40], [41]. In a similar manner, nucleic acids can
be exploited in MC-Tx to carry information. DNA has a
double-stranded structure, whereas RNA is single-stranded
molecule often folded on to itself, and the existence of
hydroxyl groups in RNA makes it less stable to hydrolysis,
compared to DNA. Hence, DNA can be expected to outper-
form RNA as an information-carrying molecule.

Recent advancements in DNA/RNA sequencing and
synthesis techniques have enabled DNA-encoded MC
[42]–[44]. DNA/RNA strands having different properties,
i.e., length [45], dumbbell hairpins [42], [46], short-
sequence motifs/labels [43], and orientation [47], can
be utilized to encode the bitwise information. For infor-
mation detection, solid-state-based nanopores [43], [48]
and DNA-origami-based nanopores [49] can be utilized to

1306 PROCEEDINGS OF THE IEEE | Vol. 107, No. 7, July 2019



Kuscu et al.: Transmitter and Receiver Architectures for MCs

Fig. 3. System model for DNA-based MCs.

distinguish the information symbols based on the proper-
ties of DNA/RNA strands by examining the ionic current
characteristics during translocation, i.e., while DNA/RNA
strands pass through the nanopores, as illustrated in Fig. 3.
The utilization of nanopores for DNA symbol detection
also enables the miniaturization of MC-capable devices
toward the realization of IoBNT. According to [42], 3-bit
barcode-coded DNA strands with dumbbell hairpins can be
detected through nanopores with 94% accuracy. In [47],
four different RNA molecules having different orientations
are translocated with more than 90% accuracy while pass-
ing through transmembrane protein nanopores. Nanopore-
based detection can also pave the way for detecting cancer
biomarkers from RNA molecules for early detection of
cancers, as in [50]. However, this requires high sensitivity
and selectivity.

Translocation time of DNA/RNA molecules through
nanopores depends on the voltage, concentration, and
length of the DNA/RNA symbols, and the translocation
of symbols can take from a few milliseconds up to 100-
ms time frames [42], [46]. Considering the slow diffusion
channel in MC, transmission/detection of DNA-encoded
symbols does not introduce a bottleneck, and multiple
detections can be performed during each symbol trans-
mission. Therefore, the utilization of DNA/RNA strands
is promising for high-capacity communication between
nanomachines, as the number of symbols in the mod-
ulation scheme can be increased by exploiting multiple
properties of DNA/RNA at the same time. Hence, the uti-
lization of DNA as an information-carrying molecule paves
the way for high-capacity links between nanomachines
by enabling a higher number of molecules that can be
selectively received [51].

In addition, information can be encoded into the base
sequences of DNAs, which is also known as nucleotide

shift keying (NSK). In [52], 35 distinct data files over
200 MB were encoded and stored by using more than
13 million nucleotides. More importantly, this paper pro-
poses a method for reading the stored data in DNA
sequences using a random access approach. Owing to the
high information density of DNA, application of NSK in
DNA-based MC may boost the typically low data rates
of MC up to the extent of competing with traditional
wireless communication standards. Hence, NSK can enable
indoor artificial molecular wireless communications with
data rates up to hundereds of megabits per second, and
this leads to a novel communication paradigm, molecular
information-fidelity (Mi-Fi). In Mi-Fi, multiuser channel
access can be achieved through molecular division multiple
access (MDMA) capability, i.e., capable of communicating
via multiple types of information-carrying molecules, e.g.,
different DNA strands, at once.

In NSK, information-carrying DNA and RNA strands can
be placed into bacteria and viruses. In [53], bacteria-based
nanonetworks, where bacteria are utilized as a carrier
of IMs, have been proposed and analytically analyzed.
In [54], a digital movie is encoded into DNA sequences,
and these strands are placed into bacteria. However,
DNA/RNA reading/writing speed and cost at the moment
limit the utilization of NSK in a practical system. Theoreti-
cal and experimental investigations of DNA-/RNA-encoded
MC stand as a significant open research issue in the MC
literature.

b) Elemental ions: Elemental ion concentrations, e.g.,
Na+, K+, and Ca2+, dictate many processes in biological
systems. For instance, in a neuron at steady state, more
Na+ and K+ ions are maintained via active transmem-
brane ion channels in extracellular and intracellular media,
respectively, and an action potential is instigated by a
surge of Na+ ions into the cell upon the activation of

Vol. 107, No. 7, July 2019 | PROCEEDINGS OF THE IEEE 1307



Kuscu et al.: Transmitter and Receiver Architectures for MCs

transmembrane receptors by neurotransmitters from other
neurons. On the other hand, Ca2+ ions are utilized in
many complex signaling mechanisms in biology, including
neuronal transmission in an excitatory synapse, exocytosis,
cellular motility, apoptosis, and transcription [55]. This
renders elemental ions one of the viable ways to com-
municate with living organisms for a diverse range of
possible applications, such as neural interfacing, disease
monitoring, diagnosis, and treatment. Correspondingly,
many transmitter device [56] and nanonetwork architec-
tures [57] based on the elemental ion signaling have been
proposed in the literature.

c) Neurotransmitters: Neural interfacing is one of the
hottest contemporary research topics with applications,
including neural prosthetics, spinal cord injury treatment,
and brain–machine interfaces. In this respect, stimulation
of neurons using their own language, i.e., neurotransmit-
ters, stands out as the best practice. Various transmit-
ter architectures for the most common neurotransmitters,
e.g., glutamate, γ-aminobutyric acid (GABA), aspartate,
and acetylcholine, have been reported in the literature
[32], [58], [59].

d) Proteins: Proteins comprise the basic building
blocks of all mechanisms in life. They are synthesized
by cells from amino acids utilizing information encoded
inside DNA and regulate nearly all processes within the
cell, including the protein synthesis process itself so that
they form a self-regulatory network. Proteins are com-
monly used as IMs by biological systems both in intracel-
lular pathways, e.g., enzymes within vesicles between the
endoplasmic reticulum and the Golgi apparatus [60], and
intercellular pathways, e.g., hormones within exosomes,
vesicles secreted from a multitude of cell types via exo-
cytosis [61]. Artificial transmission of proteins has been
long considered within the context of applications, such as
protein therapy [62]. From MC point of view, use of pro-
teins as IMs by genetically engineered bacteria is regarded
as one of the possible biological MC architectures, where
engineered genetic circuits have been proposed to imple-
ment various logic gates and operations necessary for
networking [63].

e) Other molecules: Many other types of molecules
have been used or considered as IMs in the literature.
These include synthetic pharmaceuticals [64], therapeutic
nanoparticles [65], as well as organic hydrofluorocarbons
[66] and isomers [67].

2) Nanomaterial-Based MC-Tx Architectures: In the liter-
ature, MC-Tx is generally assumed to be an ideal point
source capable of perfectly transmitting molecular mes-
sages encoded in the number, type, or release time of
molecules to the channel instantly or continuously, neglect-
ing the stochasticity in the molecule generation process
and the effect of the Tx geometry and channel feed-
back. Despite various studies investigating MC, the phys-
ical implementation of MC-Tx stands as an important
open research issue, especially in microscale/nanoscale.

However, there exist some works on the macroscale
demonstration of MC. Farsad et al. [68] implemented a
macroscale MC system with an electronically controlled
spray as MC-Tx, which is capable of releasing alcohol,
and alcohol metal-oxide sensor as MC-Rx. According to
this experiment, the macroscale MC setup achieves 0.2 b/s
with 2-m communication range. Since there is almost no
microscale/nanoscale implementation of MC-Tx, we inves-
tigate and propose MC-Tx architectures by exploiting
recent advancements in nanotechnology, novel materials,
and microfluidics.

As discussed in Section II-A, design of MC-Tx in
microscale/nanoscale is an extremely challenging task,
including various requirements, such as biocompability,
miniaturization, and lifetime of the device. For reducing
the dimensions of MC-Tx architectures to microscales,
microfluidics and microfluidic droplet technologies are
promising. Inside droplets, IMs can be transmitted in a
precisely controlled manner, such that even logic gates can
be implemented with microfluidic chips, as demonstrated
in [69]. Feasibility of utilizing droplets for communica-
tion purposes has been suggested in [70]. In addition,
microfluidic chips can be fabricated by using polydimethyl-
siloxane (PDMS) that is a biocompatible polymer [71].
Farsad et al. [72] theoretically compared the achievable
data rates of passive transport, i.e., diffusive channel, and
active transport, i.e., flow-assisted channel via external
pressure or molecular motors, in a microfluidic environ-
ment. According to this study, active transport improves
achievable data rates, owing to faster movement of IMs
from Tx to Rx compared to passive transport. A new mod-
ulation scheme based on the distance between droplets
has been introduced in [73] by exploiting hydrodynamic
microfluidic effects.

Although it is not originally proposed as MC-Tx,
microfluidic neural interfaces with chemical stimulation
capabilities, i.e., devices that release neurotransmitters,
such as glutamate or GABA to stimulate or inhibit neural
signals [74]–[76], operates same as MC-Tx. Therefore,
the studies on neural interfaces with chemical stimula-
tion capabilities can be considered as the baseline for
designing microscale/nanoscale MC-Tx. However, leakage
of IM molecules, while there is no signal transmission,
stands as a significant challenge for microfluidic-based
MC-Txs. It is not possible to eliminate the problem, but
there are some possible solutions to reduce or control
the amount of leakage. Hydrophobic nanopores can be
utilized, as shown in Fig. 4(a), such that the liquid inside
the container can be separated from the medium when
there is no pressure inside the MC-Tx. This solution has
two drawbacks. Since there is a need for external pressure
source, it is hard to design a practical stand-alone MC-Tx
with this setup. Second, this solution does not completely
eliminate the IM leakage; hence, Jones and Stelzle [77]
suggest the utilization of porous membranes and electrical
control of fluids to further improve MC-Tx against IM
leakage.

1308 PROCEEDINGS OF THE IEEE | Vol. 107, No. 7, July 2019



Kuscu et al.: Transmitter and Receiver Architectures for MCs

Fig. 4. MC-Tx architectures. (a) Microfluidic-based MC-Tx with hydrophobic nanopore. (b) Microfluidic-based MC-Tx with hydrophobic

nanopore and nanoporous graphene membrane. (c) MC-Tx with thin film hydrogels with nanoporous graphene membrane. (d) MC-Tx with

molecule wax and nanoporous graphene membrane.

Nanoporous graphene membranes can provide molecule
selectivity [78] by adjusting the pore size depending on
the size of IM. In addition, graphene is also proven to be
biocompatible, as demonstrated in [79]. If the pore size of
the graphene membrane can be adjusted in a way that IMs
can barely pass through, negatively charging the graphene
membrane can decrease the pore size with the additional
electrons, such that some level of control can be introduced
to reduce the IM leakage. In addition, an electrode plate
placed at the bottom of the container can be utilized to
pull and push charged IMs, as illustrated in Fig. 4(b). This
way, an electric field (E-field) can be generated in the
liquid containing charged IMs, and the direction of E-field
can be utilized to ease or harden the release of charged
IMs. Thanks to their biocompatibility and controlled IM
release mechanism, microfluidic-based MC-Txs enhanced
with nanoporous membranes pave the way for several in-
body applications, e.g., enabling communication between
nanomachines flowing through the human body, inter-
facing with cells, and implementing artificial synapses,
toward realizing IoBNT.

MC-Tx can be also realized with electrical stimuli-
responsive thin film hydrogels by performing E-field-
modulated release and uptake of IMs [80], as shown
in Fig. 4(c). Hydrogels, widely utilized in smart drug deliv-
ery, are biocompatible polymers that can host molecules
and reversibly swell/deswell in a water solution upon the
application of the stimuli, resulting in release/uptake of
molecules. This architecture can be further improved via
porous graphene controlling its microenvironment. The
E-field stimuli can be generated by two electrodes placed
at the opposite walls of the reservoir. Refilling the IM
reservoirs is another significant challenge in the MC-Tx
design, which can be solved with the replenishable drug
delivery methods, e.g., oligodeoxynucleotides (ODN) mod-
ification [81], click chemistry [82], and refill lines [83].
In this way, the utilization of hydrogels further enhances
in-body MC applications by extending the device lifetimes
via molecule uptake mechanisms.

Up to this point, we consider the design of MC-Tx only
in the liquid medium. For airborne MC, we propose an MC-
Tx architecture consisting of a molecular reservoir sealed
by a porous graphene membrane, accommodating a wax

layer containing IMs, as shown in Fig. 4(d). The rest of the
reservoir is filled with water, in which the IMs dissolve.
Upon the application of heat via E-field through con-
ducting walls of the reservoir, the module sweats IM-rich
water through membrane pores, and IMs become airborne
upon evaporation and provide a continuous release of IMs.
For this reason, this MC-Tx architecture is promising for
applications, where the continuous release of molecules is
favorable in order to increase the detection probability and
reliability, such as sending molecular warning signals to
plants and insects.

3) Biological MC-Tx Architectures: An alternative
approach to nanomaterial-based architectures for MC,
which, in most cases, are inspired by their biological
counterparts, resides in rewiring the already established
molecular machinery of the biological realm to engineer
biological nanomachines that network via MC to accom-
plish specific tasks. At the cost of increased complexity,
this approach has various advantages over nanomaterial-
based designs, including inherent biocompatibility and
already integrated production, transport, and transmission
modules for a wealthy selection of IMs and architectures.

a) Bacterial conjugation-based transmission: Bacterial
conjugation is one of the lateral gene transfer processes
between two bacteria. More specifically, some plasmids
inside bacteria, mostly circular small double-stranded DNA
molecules physically distinct from the main bacterial DNA,
can replicate and transfer itself into a new bacteria [84].
These plasmids encode the conjugative “sex” pilus that,
upon receiving a right molecular stimulus from a neigh-
boring bacteria, is translated. The produced pilus extends
out of the donor cell, attaches to the recipient cell, and,
then, retracts to get the two cells in contact with their
intracellular media joined through the pilus hole. Single-
strand DNA (ssDNA) of the plasmid is then transferred
from the donor cell to the recipient cell. Utilizing bacterial
conjugation as a means of information transmission for MC
necessitates at least partial control over bacteria behavior,
which is achieved by means of genetically engineering the
bacteria.

At the core of all genetical engineering schemes lies
the concept of gene regulation via a process known as

Vol. 107, No. 7, July 2019 | PROCEEDINGS OF THE IEEE 1309



Kuscu et al.: Transmitter and Receiver Architectures for MCs

RNA interference (RNAi), which provided us the means
of manipulating gene expressions in targeted cells or bac-
teria, paving the way for genetically engineered bacteria-
based MC architectures [85]. RNAi is observed to serve
as a mediator of interkingdom MC [86]. In particular,
it is shown that short hairpin RNA (shRNA) express-
ing bacteria elicit RNAi in mammals [87], rendering
bacteria-mediated RNAi-based diagnosis and therapeu-
tics a promising prospect [88]. In this respect, recently,
McKay et al. [89] proposed a platform of genetically engi-
neered bacteria as vehicles for localized delivery of ther-
apeutics toward applications for Crohn’s disease. From
MC perspective, genetically engineered bacteria have been
envisioned to be utilized in various ways with different
transmitter architectures. In the following, we collate var-
ious novel research directions in biological transmitter
architectures for MC enabled by the recent advancements
in genetic engineering.

It is important to note that conjugative pili are typically
few microns in length, which, from the MC point of view,
dramatically decreases the transmission radius. Accord-
ingly, [53] has proposed engineered bacteria with flagella,
e.g., E. coli as the carriers of information, i.e., sequenced
DNA strands, which establishes a communication link
between two nanomachine nodes that can interface and
exchange DNA strands with the bacteria. The sequenced
DNA message resides within a plasmid. The behavior of
the bacteria is controlled via chemotaxis by the release
of attractants from the nodes and regulated by encoded
active regions on the plasmid. To avoid interference with
behavior regulation, the message section of the plas-
mid is inactivated, i.e., it is not expressed, which can
be achieved by avoiding consensus promoter sequences
within the message. Consensus promoters are necessary
for the RNA polymerase to attach to the DNA and start
transcription. The message section of the plasmid also
contains the destination address, which renders message
relaying across nodes possible. The authors develop a sim-
ulator for the flagellated bacteria propagation, which they
combine with analytical models for biological processes
involved to obtain the end-to-end delay and capacity of
the proposed MC channel for model networking tasks.
In [90], utilization of flagellated bacteria for the medium-
range (μm–mm) MC networks is suggested, where the
authors present a physical channel characterization of the
setup together with a simulator based on it. Sugrañes and
Akyildiz [91] extend the model in [53] by accounting for
mutations in the bacteria population and also considering
the asynchronous mode of operation of the nodes. Bala-
subramaniam et al. [92] also considered a similar setup
utilizing bacterial conjugation but employ opportunistic
routing, where opportunistic conjugation between bacteria
is allowed in contrast to strict bacteria-node conjuga-
tion assumed in [53]. However, it requires node labeling
of bacteria and additional attractant release by bacteria
to facilitate the bacteria–bacteria contact for opportunis-
tic routing. The authors extend this work with [93] by

additionally assuming the nanomachine nodes as capable
of releasing antibiotics that kill bacteria with useless or no
content to decrease the noise levels by avoiding overpopu-
lation. Moreover, they allow a multiple number of plasmids
per bacterium, which, contrary to their previous work and
[53], allows simultaneous communication between multi-
ple source and destination nodes over the same network.

An important factor that restricts reliable successful
delivery in conjugation-based bacterial networks is incom-
plete DNA transfer due to the fragile nature of the process,
the effect of which is pronounced over multiple transfers.
To mitigate losses due to this effect, which always results
in a loss of information from the tail part of DNA, [94]
devises a forward–reverse coding (FRC) scheme, where
messages are encoded in both directions and sent simul-
taneously over the same channel via dedicated sets of
bacteria, to equally distribute losses to both ends of DNA.
The performance of FRC is later compared with a cyclic
shift coding (CSC) scheme, in which a DNA message is
partitioned into N smaller blocks and is cyclically shifted
to create N versions of the same content starting with
corresponding blocks, and similar to FRC transmitted
simultaneously via dedicated bacteria populations [95].
Expectedly, CSC provided higher link probability than FRC,
which outperformed straight encoding.

The MC-Tx module of the architecture based on the
bacterial conjugation outlined earlier, indeed, is composed
of several submodules, i.e., the transmitter module of
the nanomachine node, the bacteria itself via chemotaxis,
and the pilus-based sexual conjugation module, which is
a reflection of the trademark high complexity involved
with biological architectures. In return, the reward is the
achievement of unprecedented data rates via MC, owing
to the high information density of DNA.

b) Virus-based transmission: Viruses have initially
been studied as infectious agents and tools for investigat-
ing cell biology; however, their use as templates for trans-
ferring genetic materials to cells has provided us the means
of genetically engineering bacteria [96] and unlocked a
novel treatment technique in medicine, e.g., gene therapy
[97]. A virus is a biomolecular complex that carries DNA
(or RNA), which is packed inside a protein shell, called
capsid, that is enveloped by a lipid layer. The capsid has
(at least) an entrance to its interior, through which the
nucleic payload is packed via a ring ATPase motor protein
[98]. Located near the entrance are functional groups that
facilitate docking on a cell by acting as ligands to receptors
on it. Once docked, the nucleic content is injected into the
cell, which triggers the cell’s production line to produce
more of virus’ constituent parts and DNA. These self-
organize into fully structured viruses, which finally burst
out of the host cell to target new ones.

From MC perspective, viral vectors are considered to be
one of the possible solutions in transmitting DNA between
nanonetworking agents. The concept is similar to a bacte-
rial conjugation-based transmission, as both encode infor-
mation into transmitted DNA, but, in contrast, viral vectors

1310 PROCEEDINGS OF THE IEEE | Vol. 107, No. 7, July 2019



Kuscu et al.: Transmitter and Receiver Architectures for MCs

are immotile and their propagation is dictated by passive
diffusion. However, they are comparatively smaller than
bacteria so that more of them can be deployed in a given
channel, and they diffuse faster. Moreover, the ligand–
receptor docking mechanism, which serves as a header
for receptor-/cell-specific long-range targeting, enables the
possibility of the design of very complex and large-scale
networking schemes with applications in gene therapy [4].
Furthermore, considering the high information density of
DNA, virus-based MC stands out as one of the MC protocols
viable to support high data rates. Yet, models of MC net-
works utilizing viral vectors are still very few. In particular,
[63] proposes the utilization of engineered cells as plat-
forms for devices and sensors to interface to nanonetworks.
These engineered cells are assumed to be capable of virus
production and excretion to facilitate desired networking,
where a modular approach is presented for modeling
of the genetic circuitry involved in the modulation of
viral expression based on incoming extracellular signaling.
Based on the model developed in [63], [99], and [100]
investigate viral MC networking between nanomachines
that communicate with each other in a diffusive medium
via DNA messages transmitted within viruses, where the
former analyses the reliability of a multipath topology and
the latter derives reliability and delay in multihop relay
networks.

c) Genetic circuit-regulated protein transmission: So far,
we have investigated biological transmitter architectures,
where the message to be conveyed has been encoded in
sequences of nucleotides, e.g., DNA or RNA. Yet, the most
abundant form of intercellular MC interaction in nature
occurs via proteins, whose expression levels are deter-
mined by the genetic circuitry and metabolic state within
the cells. Typically, proteins that are produced within a
cell via transcriptional processes are either excreted out
via specialized transmembrane protein channels or packed
into vesicles and transported to the extracellular medium
via exocytosis. As a result of exocytosis, either the vesicle
is transported out wholly, e.g., an exosome [101], or it
fuses with the cell membrane and only the contents are
spewed out. Proteins on the membranes of exosomes
provide addressing via ligand–receptor interactions with
membrane proteins of the recipient cell. Upon a match,
the membranes of exosome and the recipient cell merge,
and the exosome contents enter the recipient cell. This
establishes a one-to-one MC link between two cells. In case
the contents are merely spewed out to the external
medium, they diffuse around contributing to the overall
concentrations within the extracellular medium. This cor-
responds to local message broadcast, and it has been long
observed that populations of bacteria regulate their behav-
ior according to the resulting local molecule concentrations
resulting from these broadcast messages, referred to as the
phenomenon of quorum sensing.

This mode of communication, even though lacking
the information density of nucleotide chains, therefore
supporting lower data rates, is commonly employed,

by nature, as MC schemes. They are considerably more
energy efficient compared to DNA transmission schemes,
which justifies their use, by nature, as signaling agents
for comparatively simple nanonetworking tasks. In this
respect, in the context of MC among bacterial networks,
quorum sensing has been proposed as a means to achieve
synchronization among the nodes of the network [102],
as well as a tool for power amplification of MC sig-
nals, increasing the range of transmitted signals [103].
Instead of utilizing quorum sensing to improve MC,
Martins et al. [104] proposed quorum jamming to sup-
press the intrinsic quorum networking of a bacterial net-
work with the aim of preventing them to form biofilms.
This idea has possible applications in fighting infectious
diseases caused by antibiotic-resistant bacteria, where the
jamming signals can be delivered to the infectious pop-
ulation via the use of genetically engineered bacteria.
In contrast to these works that consider quorum sensing,
[105] considers exosome secretion as a possible means for
the realization of nanonetworks composed of a large num-
ber of bionanomachines. Unluturk et al. [18] presented a
detailed model of engineered genetic circuitry based on
mass action laws that regulate gene expressions, which,
in turn, dictate transmitter protein production. This paper
stands as a basis for the future engineered genetic circuitry-
based protein transmission modeling.

d) Enzyme regulated Ca2+ circuits: Ca2+ ions are
utilized in many complex signaling mechanisms in biol-
ogy, including neuronal transmission in an excitatory
synapse, exocytosis, cellular motility, apoptosis, and tran-
scription [55]. Moreover, they play a major role in intra-
cellular and intercellular signal transduction pathways,
where the information is encoded into local Ca2+ concen-
tration waves under the control of enzymatic processes.
Intracellular organelles, including mitochondria and endo-
plasmic reticulum, accumulate excess Ca2+ ions and serve
as Ca2+ storages. Certain cellular events, such as an
extracellular signal generated by a toxin, trigger enzy-
matic processes that release bound Ca2+ from organelles,
effectively increasing local cytosolic Ca2+ concentrations
[106]. These local concentrations can propagate like waves
within cells and can be injected to adjacent cells through
transmembrane protein gap junction channels, called con-
nexins [107].

Establishing controlled MC using this inherent signal-
ing mechanism was first suggested by [108], where the
authors consider communicating information between two
nanomachines over a densely packed array of cells that
are interconnected via connexins. The authors simulate
MC within this channel to analyze the system parameter-
dependent behavior of intercellular signal propagation
and its failure. They also report on experiments relating
to so-called cell wires, where an array of gap junction
transfected cells are confined in a wire configuration and
signals along the wire are propagated via Ca2+ ions. The
communication was characterized as limited range and
slow speed. Later, Ca2+ relay signaling over one-cell-thick

Vol. 107, No. 7, July 2019 | PROCEEDINGS OF THE IEEE 1311



Kuscu et al.: Transmitter and Receiver Architectures for MCs

cell wires was investigated in [63] and [109] via dedi-
cated simulations, where the former explored amplitude
and frequency modulation characteristics and the latter
aimed at understanding the communication capacity of the
channel under stochastic effects. Simulations to determine
the effects of tissue deformation on Ca2+ propagation and
the capacity of MC between two nanomachines embedded
within 2-D cell wires with a thickness of multiple cells were
carried out in [110]. As a part of their study, the authors
propose various transmission protocols and compare their
performance in terms of achieved rates. In a later study
[111], the authors employ Ca2+ signaling nanomachines
embedded within the deformable cell arrays to infer defor-
mation status of the array as a model for tissue deforma-
tion detection. This is achieved by estimating the distance
between nanomachines from observed information metrics
coupled with strategic placements of nanomachines. Moti-
vated by tissue health inference via embedded nanoma-
chines, Barros et al. [112] identified three categories of
cells that employ Ca2+ signaling, namely, excitable, nonex-
citable, and hybrid, which, respectively, model muscle
cells, epithelium cells, and astrocytes, and model the Ca2+

communication behavior within channels that comprised
of these cells. We refer the reader to [113] for a more
detailed discussion of existing literature, theoretical mod-
els, experiments, applications, and future directions in this
field of MC.

e) Other biological architectures: In addition to the
biological transmission architectures described earlier,
molecular motors’ sliding on cytoskeletal protein struc-
tures, e.g., microtubules, and carrying cargo, i.e., vesicles,
between cells is another option that has been considered
by the MC community. In particular, Moore et al. [114]
and Enomoto et al. [115] describe a high-level architecture
design for MC over such channels. The comparison of
active, i.e., molecular motors on microtubules, and passive,
i.e., diffusion, vesicle exchanges among cells shows that
active transport is a better option for intercellular MC in
case of a low number of available vesicles and passive
transport can support higher rates when large numbers
of vesicles are available [116]. Two design options to
form a microtubule nanonetwork in a self-organizing man-
ner, i.e., via a polymerization/depolymerization process
and molecular motor-assisted organization, are proposed
in [117]. A complementary approach to molecular motor-
based microtubular MC is presented in [118], where an
on-chip MC test bed design based on kinesin molecular
motors is presented. In this approach, instead of molecular
motors gliding over microtubules as carriers of molecules,
microtubules are the carriers of molecules, e.g., ssDNA,
gliding over a kinesin covered substrate.

Other transmitter approaches that involve the use
of biological entities are biological-nanomaterial hybrid
approaches. A promising approach is to utilize IM produc-
tion mechanisms of bacteria in nanomaterial-based trans-
mitter architectures. In this direction, Sankaran et al. [119]
report on an optogenetically controlled living hydrogel,

that is, a permeable hydrogel matrix embedded with bac-
teria from an endotoxin-free E. coli strain, which releases
IMs, i.e., antimicrobial and antitumoral drug deoxyviola-
cein, in a light-regulated manner. The hydrogel matrix is
permeable to deoxyviolacein; however, it spatially restricts
the movement of the bacteria. This hybrid approach pro-
poses a solution to the reservoir problem of nanomaterial-
based architectures. Moreover, to cover a variety of IMs
simultaneously, this approach can be built upon by utilizing
many engineered bacteria and controlling their states via
external stimulus to control their molecular output [120].

III. M O L E C U L A R C O M M U N I C AT I O N
R E C E I V E R

The MC-Rx recognizes the arrival of target molecules to
its vicinity and detects the information encoded in a phys-
ical property of these molecules, such as concentration,
type, or release time. To this aim, it requires a molecular
receiver antenna that consists of a biorecognition unit
followed by a transducer unit. The biorecognition unit,
i.e., the interface with the molecular channel, holds a mole-
cular recognition event that is specifically selective to the
information-carrying molecules, e.g., it selectively reacts to
these target molecules. Then, the transducer unit generates
a processable signal, e.g., electrical or biochemical signal,
based on this molecular reaction. Finally, a processing unit
is needed to detect the transmitted information based on
the output of the molecular antenna. The interconnection
of these components in an MC-Rx is illustrated in Fig. 5
[121]. Since this structure is fundamentally different from
EM communication receivers, it is necessary to thoroughly
investigate the receiver architecture specification. To this
aim, in this section, we first discuss the requirements
of a receiver to be operable in an MC application and
the communication theoretical performance metrics that
must be taken into consideration while designing the
receiver. Then, we review the available approaches in the
physical design of MC-Rxs, which can be categorized into
two main groups: 1) biological receivers based on synthetic
gene circuits of engineered bacteria and 2) nanomaterial-
based artificial MC-Rx structures.

A. Design Requirements for MC-Rx

While designing the MC-Rx, its integrability to a mobile
nanomachine with limited computational, memory, and
energy resources, which requires to operate independently
in an MC setup, must be taken into consideration. This
dictates the following requirements for the functionality
and physical design of the receiver [121].

1) In Situ Operation: In-device processing of the molec-
ular message must be one of the specifications of the
receiver since it cannot rely on any postprocessing
of the transduced signals by an external macroscale
device or a human controller.

2) Label-Free Detection: Detection of information-
carrying molecules, i.e., IMs, must be done based

1312 PROCEEDINGS OF THE IEEE | Vol. 107, No. 7, July 2019



Kuscu et al.: Transmitter and Receiver Architectures for MCs

Fig. 5. Components of an MC-Rx.

on their intrinsic characteristics, i.e., no additional
molecular labeling procedure or preparation stage is
required.

3) Continuous Operation: MC-Rx requires to
observe the molecular channel continuously to
detect the signal encoded into concentration,
type/ratio/order, or release time of molecules. Thus,
the functionality of the molecular antenna and the
processing unit should not be interrupted. Since
receptors are needed in the biorecognition unit for
sensing target molecules, it is important to have
reusable receptors, i.e., they must return to their
initial state after signal detection to be ready for the
next channel use.

4) Energy Efficiency: Due to the limitations of nanoma-
chines, the energy usage of the MC-Rx must be
optimized. In addition, as discussed in Section II-A,
batteries may not be the feasible solutions for the
long-term activity of nanomachines, e.g., as an
implanted device. Thus, the receiver may need to be
designed with EH units to be energy self-reliance.

5) Biocompatibility and Biodurability: One of the most
important MC application areas is the life sci-
ence, e.g., it promises diagnosis and treatment
techniques for diseases caused by dysfunction of
intrabody nanonetworks, such as neurodegenera-
tive diseases [3]. These in vivo applications dictate
further requirements for the device. First, it must
not have any toxic effects on the living system.
Moreover, the device needs to be flexible, not to
cause any injury to the living cells due to mechanical
mismatches. Furthermore, there must not be any
physiological reactions between the device and the

environment, and it must not cause immunological
rejection. On the other hand, the physiological envi-
ronment should not degrade the performance of the
device with time.

6) Miniaturization: Finally, to be integrated into
a nanomachine, the MC-Rx must be built on
microscale/nanoscale components.

B. Communication Theoretical
Performance Metrics

The general performance metrics defined for EM com-
munications, e.g., signal-to-noise ratio (SNR), bit error rate
(BER), and mutual information, can also be used for MC.
However, new performance metrics are needed to fully
evaluate the functionality of an MC-Rx since molecules
are used as IMs in MC. The most important performance
metrics are summarized as follows.

1) Limit of Detection (LoD): LoD is a well-known per-
formance metric in the biosensing literature [122].
It shows the minimum molecular concentration in
the vicinity of the biosensor needed for distinguish-
ing between the existence and the absence of tar-
get molecules. Since the input signal in MC-Rxs
is a physical property of information carriers, LoD
corresponds to the sensitivity metric used for eval-
uating the performance of the EM communication
receivers, which indicates the minimum input signal
power needed to generate a specified SNR at the
output of the device.

2) Selectivity: This metric is defined based on the rel-
ative affinity of the biorecognition unit to the IMs
and interferer molecules [123]. Note that interferer
molecules can be non-IMs in the medium or other
types of IMs in case of the MC system with multiple
IMs, such as molecule shift keying (MoSK). High
selectivity, i.e., very lower probability of interferer–
receptor binding compared to target molecule–
receptor binding, is needed to uniquely detect the
IMs in the vicinity of the MC-Rx.

3) Operation Range: The biorecognition unit does not
provide an infinite range of molecular concentra-
tion detection, as it has finite receptor density. The
response of the device can be divided into two
regions: the linear operation and the saturation
region [124]. The range of molecular concentra-
tion in the vicinity of the MC-Rx that does not
lead to the saturation of receptors is called the
linear operation region. In this region, the output
of molecular antenna provides better information
about changes in the molecular concentration. Thus,
when the information is encoded into the concen-
tration of molecules, the receiver must work in the
linear operation region. However, for other encoding
mechanisms, e.g., MoSK, the receiver can work in
both regions.

Vol. 107, No. 7, July 2019 | PROCEEDINGS OF THE IEEE 1313



Kuscu et al.: Transmitter and Receiver Architectures for MCs

4) Molecular Sensitivity: In addition to the aforemen-
tioned sensitivity metric that is mapped to LoD for
MC-Rxs, a molecular sensitivity can also be defined
for an MC-Rx. This metric indicates the smallest
difference in the concentration of molecules that can
be detected by an MC-Rx [123]. It is of the utmost
importance when the information is encoded in the
molecular concentration. The metric can be defined
as the ratio of changes in the output of the molecular
receiver antenna to the changes in the molecular
concentration in the vicinity of the MC-Rx when the
device performs in the linear operation region.

5) Temporal Resolution: This metric is defined to eval-
uate the speed of the molecular concentration sam-
pling by the receiver. Since the electrical processes
are much faster than molecular processes, it is
expected that the diffusion and the binding kinetics
limit the temporal resolution. Thus, the biorecogni-
tion unit should be realized in a transport-limited
manner to detect all the messages carried by mole-
cules into the vicinity of the MC-Rx, i.e., the binding
kinetics should not be a limiting factor on the sam-
pling rate [121].

C. Physical Design of MC Receiver

Most of the existing studies on the performance of MC
ignore the physical design of the receiver and assume
that the receiver can perfectly count the number of mole-
cules that: 1) enter a reception space with transparent
boundaries; 2) hit a 3-D sphere that absorbs molecules; or
3) bind to receptors located on its surface [6]. How-
ever, the processes involved in the molecular-to-electrical
transduction affect the performance of the receiver. Thus,
the comprehensive communication theoretical modeling of
these processes is required. Available approaches in the
physical design of MC-Rxs can be categorized into two
main groups as follows.

1) Biological MC-Rx Architectures: Synthetic biology,
the engineering of biological networks inside living cells
by modifying the natural gene circuits or creating new
synthetic ones, has seen remarkable advancements in
the last decade. As a result, it becomes possible to
device engineered cells, e.g., bacteria, for use as bio-
logical machines, e.g., sensors and actuators, for various
applications. Synthetic biology also stands as a promising
means of devising nanoscale biotransceivers for IoBNT
applications, by implementing transmission and reception
functionalities within living cells [18]. Due to their nature,
the biological MC-Rxs are promising for in vivo applica-
tions, such as monitoring the condition of a living organ-
ism, e.g., human or animal, regenerating biological tissues
and organs, localized cancer treatment, immune system
support, and interfacing artificial devices with nervous
systems.

Synthetic biology is already mature enough to allow per-
forming complex digital computations, e.g., with networks

Fig. 6. Biological circuitry of an MC-Rx [127].

of genetic NAND and NOR gates, as well as analog computa-
tions, such as logarithmically linear addition, ratiometric,
and power-law computations, in synthetic cells [125]. Syn-
thetic gene networks integrating computation and memory
is also proven feasible [126]. More importantly, the tech-
nology enables implementing bionanomachines capable
of observing individual receptors, as naturally done by
living cells. Thus, it stands as a suitable domain for
practically implementing more information-efficient MC
detectors based on the binding state history of individual
receptors, as discussed in Section V.

In DNAs, gene expression is the process that produces a
functional gene product, such as a protein. The rate of this
expression can be controlled by the binding of another pro-
tein to the regulatory sequence of the gene. In biological
circuits, activation and repression mechanisms that regu-
late the gene expressions are used to connect DNA genes
together, i.e., the gene expression of a DNA generates a
protein that can then bind to the regulatory sequence of
the next DNA to control its expression [128]. In [129],
polymerases per second (PoPS) is defined as the unit
of input and output of a biological cell, i.e., the circuit
processes a PoPS signal as input and generates another
PoPS signal at its output using the aforementioned con-
nections among DNAs. The literature in applying synthetic
biology tools to design bacteria-based MC-Rxs is scarce.
An MC biotransceiver architecture integrating molecular
sensing, transmitting, receiving, and processing functions
through genetic circuits is introduced in [18]. However,
the analysis is based on the assumption of linearity and
time invariance of the gene translation networks and does
not provide any insight into the associated noise sources.
The necessary biological elements for an MC-Rx are shown

1314 PROCEEDINGS OF THE IEEE | Vol. 107, No. 7, July 2019



Kuscu et al.: Transmitter and Receiver Architectures for MCs

in Fig. 6 [127]. The process of receiving information is
initiated by the receptor activation expression, which gets
a PoPS auxiliary signal as its input and generates receptor
proteins, denoted by R, at its output. The incoming signal
molecules from the molecular channel, i.e., SRx, then bind
to these receptors in the ligand–receptor binding unit and
form activator complexes, called RS. Finally, the concen-
tration of RS initiates the output transcription activation,
in which RNA polymerase proteins, RNAP, bind to the pro-
moter sequence, PRx, and produce the PoPS output signal,
PoPSOut. Assuming that the intracellular receiver environ-
ment is chemically homogeneous, the transfer function of
the aforementioned biological circuit is derived in [127] to
provide a system-theoretic model for the MC-Rx. Genetic
circuits are shown to provide higher efficiency for analog
computations compared to digital computations [130].
Thus, analog computing functionalities of genetic circuits
are utilized in [131] to derive an analog parity-check
decoder circuit.

The major challenge in using genetic circuits for
implementing MC-Rx arises from the fact that the infor-
mation transmission in biological cells is through mole-
cules and biochemical reactions. This results in nonlinear
input–output behaviors with system-evolution-dependent
stochastic effects that are needed to be comprehen-
sively studied to evaluate the performance of the device.
In [132]–[134], initiative studies on mathematical model-
ing of cellular signaling are provided. However, it is shown
in [135] that the characterization of the communication
performance of these systems is not analytically tractable.
In addition, a computational approach is proposed to char-
acterize the information exchange in a biocircuit-based
receiver using the experimental data published in [136].
The results in [135] reveal that the rate of information
transfer through biocircuit-based systems is extremely lim-
ited by the existing noises in these systems. Thus, com-
prehensive studies are needed to characterize these noises
and derive methods to mitigate them. In addition, exist-
ing studies have only focused on the single transmitter–
receiver systems; thus, the connection among biological
elements for MC with multiple receiver cells remains as
an open issue.

2) Nanomaterial-Based Artificial MC-Rx Architectures:
MC-Rxs with artificial structures can be used in both in
vivo applications, for which biocompatibility of the device
and biostability of its response must be investigated, and
in vitro applications. Apart from biomedical applications,
such as health monitoring and drug delivery that can also
be achieved using biological MC-Rxs, artificial MC-Rxs
can also be used for applications, such as environmental
monitoring to detect paste, pollution, toxic or radioactive
agents, food and water quality control, and safe conversion
of undesired materials [15]. Similar to biosensors, MC-Rxs
are also designed to detect the concentration of an analyte
in a solution. Thus, existing literature on MC-Rx design
is focused on analyzing the suitability and performance of

available biosensing options for receiving the information
in the MC paradigm [68], [121], [137]–[139]. In this
direction, researchers must consider the fundamental dif-
ferences between a biosensor and an MC-Rx, arising from
their different application areas, as stated in the following.

1) Biosensors are designed to perform typically in
the equilibrium condition. However, MC-Rxs must
continuously observe the environment and detect
the information encoded into a physical property
related to the molecules, such as concentration,
type/ratio/order, or arrival time.

2) Biosensors are mostly designed for laboratory appli-
cations with macroscale readout devices and human
observers to compensate for the lack of an integrated
processor, which is not applicable for an MC-Rx.

Thus, while the biosensing literature provides insights for
MC-Rx design, ICT requirements of the device and its
performance in an MC paradigm must be considered to
reach appropriate solutions.

Among existing biosensing options, the electrical biosen-
sors are mainly under the focus of the MC-Rx design [121].
The remaining options, i.e., optical and mechanical sens-
ings [140], [141], need macroscale excitation and detec-
tion units, making them inappropriate for an MC-Rx that
requires in situ operation. Biocatalytic- [142] and affinity-
based [143] sensors are two types of electrical biosensors
with different molecular recognition methods. Biocatalytic
recognition is based on two steps. First, an enzyme, immo-
bilized on the device, binds with the target molecule,
producing an electroactive specie, such as hydrogen ion.
The arrival of this specie near the working electrode of the
transducer is then being sensed, as it modulates one of the
electrical characteristics of the device. Glucose and glu-
tamate sensors are examples of the biocatalytic electrical
sensors [142], [144]. Alternatively, binding of receptor–
ligand pairs on the recognition layer of the sensor is the
foundation of affinity-based sensing [143].

Affinity-based sensing, which is feasible for a wider
range of target molecules, such as receptor proteins
and aptamer/DNAs [143], [145], provides a less compli-
cated sensing scheme, compared to the biocatalytic-based
sensing. For example, in the biocatalytic-based sensing,
the impact of the additional products of the reaction
between the target molecule and enzyme on the perfor-
mance of the device and the application environment must
be analyzed thoroughly. Hence, affinity-based recognition
is more appropriate for the design of a general MC sys-
tem. However, this recognition method is not possible for
nonelectroactive information-carrying molecules, such as
glutamate and acetylcholine, which are highly important
neurotransmitters in the mammalian central nervous sys-
tem [146]. Thus, it is essential to study the design of
biocatalytic-based electrical biosensors as MC-Rx when
these types of information carriers are dictated by the
application, e.g., communicating with neurons.

Recent advances in nanotechnology led to the design
of bioFETs, providing both affinity- and biocatalytic-based

Vol. 107, No. 7, July 2019 | PROCEEDINGS OF THE IEEE 1315



Kuscu et al.: Transmitter and Receiver Architectures for MCs

Table 1 Design Options, Performance, and Applications of bioFETs

Fig. 7. Conceptual design of a bioFET-based MC-Rx.

electrical sensings with use of nanowires (NWs), nan-
otubes, organic polymers, and graphene as the transducer
unit [144], [147]–[149]. Detection of target molecules by
bioFETs is based on the modulation of transducer conduc-
tivity as a result of either affinity- or biocatalytic-based
sensings. Simple operation principles together with the
extensive literature on FETs have been established through
many years, electrical controllability of the main device
parameters, high-level integrability, and plethora of opti-
mization options for varying applications make FET-based
biosensing technology also a quite promising approach for
electrical MC-Rx. Moreover, these sensors promise label-
free, continuous, and in situ operation in nanoscale dimen-
sions. Thus, the design of an MC-Rx based on the principles
of affinity-based bioFETs is the main approach considered
in the literature [121], [137], which will be overviewed in
the rest of this section.

a) Nanoscale bioFET-based MC-Rx architectures: As
shown in Fig. 7, a bioFET consists of source and drain elec-
trodes and a transducer channel, which is functionalized
by ligand receptors in affinity-based sensors [145]. In this
type of sensors, the binding of target molecules or ana-
lytes to the ligand receptors leads to accumulation or

depletion of the carriers on the semiconductor channel.
This modulates the transducer conductivity, which, in turn,
alters the flow of current between source and drain. Thus,
by fixing the source-to-drain potential, the current flow
becomes a function of the analyte density and the number
of analyte charges. Since the ligand receptors are being
selected according to the target molecule, i.e., ligands,
bioFETs do not require any complicated postprocessing,
such as labeling of molecules. Moreover, the measurement
of source–drain current does not need any macroscale
readout unit. Thus, bioFETs can be used for direct, label-
free, continuous, and in situ sensing of the molecules in an
environment as a stand-alone device.

One of the significant advantages of bioFETs over other
electrical sensors is their wide range of design parameters.
A list of FET-based biosensors and their applications in
sensing different type of molecules is provided in Table 1.
In the following, we further describe these vast design
options.

Type of Bioreceptors: First important design parameter
arises from the type of receptors used in the biorecognition
unit. Type of receptors causes the selectivity of receiver for
a certain type of molecules that will be used as an informa-
tion carrier in the MC paradigm. Among possible receptor
types for affinity-based bioFETs, natural receptor proteins
and aptamer/DNAs are appropriate ones for an MC-Rx
since their binding to the target molecule is reversible and
their size is small enough to be used in a nanomachine
[143], [145]. As an example, the FET transducer channel
is functionalized with natural receptors, e.g., neurorecep-
tors, to detect taste in bioelectronic tongues [157] and
odorant in olfactory biosensors [158]. An advantage of
these type of receptors is their biocompatibility that makes
them suitable for in vivo applications. In addition, use of
aptamers, i.e., artificial single-stranded DNAs and RNAs,
in the recognition unit of bioFETs provides detectors for
a wide range of targets, such as small molecules, pro-
teins, ions, aminoacids, and other oligonucleotides [148],
[156], [159]. Since an immense number of aptamer-ligand
combinations with different affinities exist, it provides a

1316 PROCEEDINGS OF THE IEEE | Vol. 107, No. 7, July 2019



Kuscu et al.: Transmitter and Receiver Architectures for MCs

powerful design option to control the selectivity of the
MC-Rx. The appropriate aptamer for a target ligand can
be found using the SELEX process, i.e., searching a large
library of DNAs and RNAs to determine a convenient
nucleic acid sequence [160].

Material Used for Transducer Channel: One of the most
important bioFET design parameters is the material used
as the transducer channel between source and drain
electrodes, which determines the receiver geometry and
affects the electrical noise characteristics of the device.
NWs [152], single-walled carbon nanotubes (SWCNTs),
graphene [161], molybdenum disulfide (MoS2) [147], and
organic materials, such as conducting polymers [162],
are some examples of nanomaterials suitable for use in
a bioFET channel. In the first generation of bioFETs, 1-D
materials, such as SWCNT and NW, were used as the
channel in a bulk form. However, using in the form of a
single material or aligned arrays outperformed the bulk
channels in terms of sensitivity and reduced noise [163].
Among possible NW materials, such as SnO2, ZnO, and
In2O3 [164], silicon NW (SiNW) bioFETs have shown
high sensitivity, high integration density, high-speed sam-
pling, and low power consumption [165]–[167]. How-
ever, their reliable and cost-effective fabrication is still an
important open challenge [148], [168]. Comprehensive
reviews exist on the performance of SiNW bioFETs, their
functionality in biomedical applications, such as disease
diagnostics, their top–down and bottom–up fabrication
paradigms, and integration within complementary metal–
oxide–semiconductor (CMOS) technology [163], [169],
[170]. It is concluded that SiNW bioFETs must be designed
according to the requirements arose by applications since
defining an ideal characteristic for these devices is difficult.
In addition, both theoretical and experimental studies are
needed to find the impact of structure parameters on
their functionality. Moreover, fine balancing of important
structural factors, such as number of NWs and their dop-
ing concentration and length, in SiNW bioFETs design
and fabrication remains a challenge, which affects the
sensitivity, reliability, and stability of the device. SWCNT-
based bioFETs offer higher detection sensitivity due to
their electrical characteristic; however, these devices also
face fabrication challenges, such that their defect-free
fabrication is the most challenging among all candidates
[171]. Note that the existence of defects can adversely
affect the performance of SWCNT bioFETs in an MC-Rx.
Moreover, for in vivo applications, the biocompatibility
of CNTs and biodurability of functionalized SWCNTs are
still under doubt [172]–[174]. While both NWs and CNTs
have 1-D structure, use of 2-D materials as the transducer
channel leads to higher sensitivity; since a planar structure
provides higher spatial coverage, more bioreceptors can be
functionalized to its surface and all of its surface atoms can
closely interact with the bond molecules. Thus, graphene,
with its extraordinary electrical, mechanical, and chemical
characteristics, is a promising alternative for the trans-
ducer channel of bioFETs [168], [175].

Fig. 8. Block diagram of a bioFET-based MC-Rx.

The intrinsic flexibility of graphene provides a higher
chance of integration into devices with nonplanar surfaces,
which can be more suitable for the design of nanomachines
in an MC application, such as communicating with neu-
rons [161]. There is currently a tremendous amount of
interest in building different configurations of graphene
bioFETs, e.g., back-gated [176] or solution-gated [175].
Moreover, researches has shown its superior sensing per-
formance for various analytes, e.g., antigens [177], DNA
[178], bacteria [179], odorant compounds [175], and
glucose [180].

General block diagram of a bioFET-based MC-Rx is
shown in Fig. 8. The biorecognition unit is the interface
between the communication channel and the receiver,
thus, it models sensing of the concentration of ligands.
The random motion of ligands near the surface of the
receiver, which is governed by the Brownian motion, leads
to fluctuations in the number of bound receptors. This
fluctuation can be modeled as a binding noise, which
depends on the transmitted signal and can adversely affect
the detection of the ligands concentration [181]. The
communication theoretical models of binding noise are
presented in [182] and [183]. Background noise, also
called biological interference [121], is resulted from the
binding of molecules different from targeted ligands that
might exist in the communication channel and show a
similar affinity for the receptors [181]. Note that this is
different from ISI and cochannel interference studied in
[184] and [185]. Stochastic binding of ligands to the
receptors modulated the conductance of the FET channel,
which is modeled by the transducer unit in Fig. 8. These
conductance changes are then reflected into the current
flowing between the source and drain electrodes of FET by
the output unit. The surface potential of the FET channel—
thus, its source-to-drain current—can be affected by unde-
sirable ionic adsorptions in application with ionic solu-
tions [121]. Hence, a reference electrode can be used
in the solution to stabilize the surface potential [123].
Finally, the transducing noise shown in Fig. 8 covers the
impact of the noise added to the received signal during
transducing operation, including thermal noise, caused
by thermal fluctuations of charge carriers on the bound
ligands, and 1/f (flicker) noise, resulted from traps and
defects in the FET channel [186]. Flicker noise can be the
dominant noise source in low frequencies, as it increases
with decreasing the frequency [121]. Detailed information
on the impact of the aforementioned noise processes on the
performance of bioFETs can be found in [186] and [187].
Moreover, experimental studies are provided in [188] on

Vol. 107, No. 7, July 2019 | PROCEEDINGS OF THE IEEE 1317



Kuscu et al.: Transmitter and Receiver Architectures for MCs

the noise resulted from the ion dynamic processes related
to ligand–receptor binding events of a liquid-gated SiNW
array FET by measuring the noise spectra of the device
before and after binding of target molecules.

While the existing biosensing literature can provide
insight for the MC-Rx design, there is a need for investi-
gation of design options according to the communication
theoretical requirements of an MC-Rx. Few studies have
focused on evaluating the performance of bioFETs in an
MC paradigm. A SiNW bioFET-based MC-Rx is modeled
in [121] based on the equilibrium assumption for the
receptor–ligand reaction at the receiver surface. The study
provides a circuit model for the transducer unit of the
receiver. This paper is further extended in [137], where
the spatial and temporal correlation effects resulting from
the finite-rate transport of ligands to the stochastic ligand–
receptor binding process are considered to derive the
receiver model and its noise statistics. In [189], an MC-Rx
consisting of an aerosol sampler, a SiNW bioFET function-
alized with antibodies, and a detection stage is designed
for virus detection. The performance of the receiver is
studied by considering the system in steady state. While
the receiver model in [189] takes into account the flicker
noise and the thermal noise, it neglects the interference
noise by assuming that the MC-Rx performs in a perfectly
sanitized room. Moreover, the models used in all of these
studies assume the ligand–receptor binding process in the
thermal equilibrium, and they do not capture well the
correlations resulting from the time-varying ligand concen-
tration occurring in the case of MCs. More importantly,
these studies only cover SiNW bioFET receivers and do
not provide much insight into the performance of other
nanomaterials as the transducer channel, such as graphene
that promises to provide higher detection sensitivity due to
its 2-D structure.

Thus, the literature misses stochastic models for
nanomaterial-based MC nanoreceiver architectures that
are needed to study the performance of the receiver in
MC scenarios, i.e., when the device is exposed to time-
varying concentration signals of different types and ampli-
tudes. These models must capture the impacts of receiver
geometry, its operation voltage characteristics, and all fun-
damental processes involved in sensing of molecular con-
centration, such as molecular transport, ligand–receptor
binding kinetics, and molecular-to-electrical transduction
by changes in the conductance of the channel. To provide
such a model for graphene-based bioFETs, the major fac-
tors that influence the graphene properties must be taken
into account. The number of layers is the most dominant
factor since electronic band structure, which has a direct
impact on the electrical properties of the device, is more
complex for graphene with more number of layers [190].
Next important parameter is the substrate used in the
graphene-based bioFET, especially when the number of
layers is less than three [190], [191]. The carrier mobility
in the graphene sheet is reported to be reduced by more
than an order of magnitude on the SiO2 substrate due

to charged impurities in the substrate and remote inter-
facial phonon scattering [192]. On the other hand, it is
shown that the impacts of the substrate can be reduced
in the suspended single-layer graphene sheet, resulting
in higher carrier mobility [193]. Moreover, as a result of
graphene’s large surface area, the impact of impurities on
its performance can be substantial [191]. The atomic type,
amounts, and functional groups on the edges of graphene,
which are hard to measure and control, are also among
the properties that can result in trial-to-trial variations
in the performance of the fabricated device [194]–[196].
In addition, inherent rippling in graphene sheet, defects,
and size of the sheet also affect the properties of the device
[191], [197]–[199].

b) Other MC-Rx architectures: Few studies exist on
the practical MC systems, taking into account the physi-
cal design of the receiver. In [68], the isopropyl alcohol
(rubbing alcohol) is used as the information carrier, and
commercially available metal-oxide semiconductor alcohol
sensors are used as MC-Rx. This paper provides a test
bed for MC with macroscale dimensions, which is later
on utilized in [200] to estimate its combined channel and
receiver model. This test bed is extended to a molecular
multiple-input–multiple-output (MIMO) system in [201]
to improve the achievable data rate. In [138], the informa-
tion is encoded in the pH level of the transmitted fluid, and
a pH probe sensor is used as the MC-Rx. Since the use of
acids and bases for information transmission can adversely
affect the other processes in the application environment,
such as in the body, magnetic nanoparticles (MNs) are
used as information-carrying molecules in microfluidic
channels in [139]. In this study, a bulky susceptometer is
used to detect the concentration of MNs and decode the
transmitted messages. In addition, the performance of MN-
based MC, where an external magnetic field is employed to
attract the MNs to a passive receiver, is analyzed in [202].

However, the focus of the aforementioned studies is
using macroscale and commercially available sensors as
the receiver. Thus, these studies do not contribute to the
design of a nanoscale MC-Rx. As discussed in Section II-B1,
recent advancements in DNA/RNA sequencing and syn-
thesis techniques have enabled DNA/RNA-encoded MC
[42], [43]. For information transmission, communication
symbols can be realized with DNA/RNA strands hav-
ing different properties, i.e., length [45], dumbbell hair-
pins [42], [46], and short-sequence motifs/labels [43].
For information detection, solid-state nanopores [43] and
DNA-origami-based nanopores [49] can be utilized to dis-
tinguish the information symbols, i.e., the properties of
DNA/RNA strands by examining the current characteristics
while DNA/RNA strands passing through the nanopores.
As presented in Fig. 3, MC-Rx contains receptor nanopores,
through which these negatively charged DNA strands pass,
owing to the applied potential, and as they do, they
obstruct ionic currents that normally flowthrough. The
duration of the current obstruction is proportional to the
length of the DNA strand that passes through, which is

1318 PROCEEDINGS OF THE IEEE | Vol. 107, No. 7, July 2019



Kuscu et al.: Transmitter and Receiver Architectures for MCs

Table 2 Comparison Matrix for MC Modulation Schemes

utilized for selective sensing. The use of nanopores for
DNA/RNA symbol detection also enables the miniaturiza-
tion of MC capable devices toward the realization of IoBNT.
According to [42], 3-bit barcode-coded DNA strands with
dumbbell hairpins can be detected through nanopores with
94% accuracy. In [47], four different RNA molecules hav-
ing different orientations are translocated with more than
90% accuracy while passing through the transmembrane
protein nanopores. The time of the translocation event
depends on the voltage, concentration, and length of the
DNA symbols, and the translocation of symbols can take up
to a few milliseconds up to 100-ms time frames [42], [46].
Considering the slow diffusion channel in MC, transmis-
sion/detection of DNA/RNA-encoded symbols does not
introduce a bottleneck, and multiple detections can be
performed during each symbol transmission.

IV. M O D U L AT I O N A N D C O D I N G
T E C H N I Q U E S F O R M C

A. Modulation Techniques for MC

In MC, several modulation schemes have been proposed
to encode information into concentration, molecule type,
and molecule release time, as shown in Table 2. The first
and simplest modulation method that was proposed for
MC is OOK, in which a certain number of molecules are
released for the high logic and no molecule is released to
represent the low logic [203]. In a similar manner, by using
a single molecule, concentration shift keying (CSK) that is
analogous to amplitude shift keying (ASK) in traditional
wireless channels is introduced in order to increase the
number of symbols in the modulation scheme by encod-
ing information into concentration levels [66]. In [204],
a similar modulation scheme based on the concentration
levels is proposed and named pulse-amplitude modulation
(PAM), which uses pulses of continuous IM release instead
of instantaneous release as in CSK. The number of concen-
tration levels that can be exploited significantly depends on
the molecule type and the channel characteristics, as ISI at
MC-Rx can be a limiting factor.

Hitherto, we have discussed modulation techniques that
use a single type of molecules. However, information can
be encoded by using multiple molecules, such that each
molecule represents different symbols, and k information
symbols can be represented with 2k different types of

molecules in MoSK [9], that is, 1-bit MoSK requires two
molecules to encode bit-0 with molecule A and bit-1 with
molecule B. The modulation of each molecule in MoSK is
based on other modulation techniques, such as OOK. MoSK
can achieve higher capacity, but the main limiting factor for
this modulation type is the number of molecules that can
be selectively received. Kabir et al. [205] further improved
MoSK by representing the low logic with no molecule
release and enabling simultaneous release of different type
of molecules, such that 2k symbols can be represented with
k molecules, which is named depleted MoSK (D-MoSK),
that is, 1-bit D-MoSK is equivalent to OOK. 2-bit D-MoSK
requires only two molecules, and four distinct symbols
can be encoded with these molecules (molecule A and
molecule B), such as N (00), A (01), B (10), and AB (11),
where N represents no molecule release. Furthermore,
Kim and Chae [67] propose the utilization of isomers,
i.e., the molecules having the same atoms in a different
orientation, and a new modulation scheme, named isomer-
based ratio shift keying (IRSK), in which information is
encoded into the ratio of isomers, i.e., molecule ratio
keying.

The release time of molecules can be also used to encode
information. Garralda et al. [204] proposed pulse position
modulation, in which the signaling period is divided into
two blocks, such that a pulse in the first block means high
logic and a pulse in the second block means low logic. More
complex modulation schemes based on release timing,
i.e., release time shift keying (RTSK), where information is
encoded into the time interval between molecule release,
have been investigated in [206] and [207]. The channel
characteristics in case of RTSK is significantly different
than other modulation schemes, as additive noise is distrib-
uted with the inverse Gaussian distribution in the presence
of flow in the channel [207] and the Levy distribution
without any flow [206].

In MC, ISI is an important performance-degrading factor
during detection due to the random motion of particles
in the diffusive channels. The effects of ISI can be com-
pensated by considering ISI-robust modulation schemes.
In [210], an adaptive modulation technique exploiting the
memory of the channel is utilized to encode information
into the emission rate of IMs, and this approach makes the
channel more robust against ISI by adaptive control of the
number of released molecules. In addition, the order of

Vol. 107, No. 7, July 2019 | PROCEEDINGS OF THE IEEE 1319



Kuscu et al.: Transmitter and Receiver Architectures for MCs

Table 3 Comparison Matrix for MC Channel Codes

molecules can be also used for information encoding as
in molecular array-based communication (MARCO) [209].
In this approach, different types of molecules are released
consecutively to transmit symbols, and by assuming perfect
molecular selectivity at the transmitter, the effects of ISI
can be reduced.

The modulation schemes based on the concentration,
molecule type/ratio/order, and molecule release time offer
a limited number of symbols. Therefore, MC suffers from
low data rates by considering limited number of symbols
and slow diffusive propagation. To tackle this problem,
a large amount of information can be encoded into the
base sequences of DNAs, i.e., NSK. For this purpose, infor-
mation can be encoded directly using nucleotides with an
error coding algorithm, such as Reed–Solomon (RS) block
codes [211], or an alphabet can be generated out of DNA
sequences (1–150 bp/letter) to encode information. The
latter approach can yield higher performance in terms of
BER, considering the complexity and the size of MC-Tx
and MC-Rx architectures. Although identification of base
pairs with nanopores can be performed with relatively low
costs and high speeds [212], [213], there is yet no practical
system to write DNA sequences with a microscale device.
Therefore, future technological advancements toward low-
cost and practical synthesis/sequencing of DNA are imper-
ative for MC communications with high data rates, e.g.,
on the order of megabits per second.

B. Coding Techniques for MC

Encoding of information before transmission is classi-
cally done for two reasons: source coding is done for statis-
tically efficient representation of data form a discrete input
source and channel coding is done to control errors that
occur due to channel noise via introducing redundant bits.
Source coding practices are independent of channel char-
acteristics, and as a result, they do not differ for MC with
respect to traditional communications from an ICT point
of view. For this reason, we do not cover source coding in
this review. On the other hand, MC channels are typically
diffusive, a process which has slow and omnidirectional
propagation. As a consequence, IMs quickly accumulate
in the channel after a series of transmissions, rendering
MC extremely noisy and susceptible to ISI. Moreover,

as coding has a computational burden on both transmitter
and receiver ends, and energy is a scarce resource at
nanoscale, energy efficiency of employed channel codes is
also a crucially important aspect. This calls for utilization
of lower complexity block codes, such as simple parity
codes or cyclic codes, e.g., the Hamming codes [214],
instead of the state-of-the-art high complexity codes with
high computational burden, such as the Turbo codes [215].
On the other hand, the noisy nature of MC channels and
overpronounced effects of ISI renders channel codes devel-
oped for conventional EM communications ill-adjusted for
MC, calling for the invention of novel coding techniques
specifically tailored for MC. Table 3 enlists the channel
coding practices so far employed in the MC literature.
In the following, we summarize these works and highlight
their contributions.

The first MC specific code in the literature is aimed
at mitigating the effects of ISI in MC, as it is the main
source of high BERs. Shih et al. [216] introduced the ISI-
free coding scheme under the MoSK modulation, where
two distinguishable molecules encode for bit-0 and bit-1,
respectively. The receiver is absorbing, i.e., detects every-
thing that hits, and it immediately receives bit-0 or bit-1
upon detection depending on the type of detected mole-
cule. The authors work with the example of a (4, 2, 1)

ISI-free code, where an (n, k, l) ISI-free code is an (n, k)

block code, i.e., maps k-bit information into n-bit code-
words, and is error-free, provided that there are no more
than level-l crossovers. Here, crossover is the phenomenon
of late detection of a molecule belonging to previously
transmitted symbols, and a level-l crossover means that
the detected molecule was transmitted one symbol ago.
The ISI-free code is a fixed code, in which it is invariant
with respect to the change in channel parameters. The
(4,2,1) code implemented in [216] is based on the idea
of finding a codebook with codewords, whose level-1
permutation sets, i.e., possible detection sequences under
maximum level-1 crossover assumption, are disjoint. How-
ever, level-1 permutation sets of codewords depend on
values of neighboring bits at the boundary of contiguous
codewords, where if they are same, crossovers between
contiguous codewords do not contribute new elements
to the level-1 permutation set, making finding codewords
with disjoint permutation sets easier. To achieve this,

1320 PROCEEDINGS OF THE IEEE | Vol. 107, No. 7, July 2019



Kuscu et al.: Transmitter and Receiver Architectures for MCs

Fig. 9. (a) Employed codewords for two states, e.g., starting with

0 or 1, together with their level-1 permutations. (b) State transition

diagram. (c) Encoding example for the ISI-free (4,2,1) code. Note the

same bits at the boundaries of contiguous words [216].

the authors devise a two-state encoder architecture, whose
codeword assignments and state diagram are illustrated
in Fig. 9 together with an example encoding. Note that
contiguous codewords always have the same neighboring
bits. The receiver decodes the information bits from the
codeword received by adding the number of 1’s modulo
n = 4 and converting the result to binary, which is a
fairly simple decoding rule, therefore favorable for MC.
The authors also compare the ISI-free (4,2,1) code with
convolutional and repetition codes and verify the com-
parable BER performance with much less computational
burden. In their work [217], the authors extend the ISI-
free (n, k, l) codes to account for higher level crossovers,
namely, for levels l = 2,3,4,5. Furthermore, they introduce
the ISI-free (n, k, l, s) codes, in which the codewords have
at least l final and s initial identical bits, instead of the
symmetric at least l identical bits at both ends of ISI-free
(n, k, l) codes, and show that they significantly outperform
the (n, k, l) codes under similar computational burdens.
In essence, the motivation for (n, k, l, s) codes comes from
the asymmetry of intercodeword error probabilities arising
from crossover at the start and at the end of a codeword.
More specifically, if one compares the probability of a
given molecule having level-l crossover forward, that is,
arriving later than the l molecules released after it, to the
probability of having level-l crossover backward, that is,
arriving earlier than the preceding l molecules, one finds
latter to fall far more rapidly with increasing l. Thus,
to reduce the computational burden, s is typically chosen
lower than l, signifying the low probability of backward
crossover errors. Finally, the authors demonstrate that ISI-
free (n, k, l, s) codes can deliver better BERs than convo-
lutional codes with less computational resources. As its

weaknesses, the work considers a very simple 1-D diffusive
channel model with positive drift velocity, which lacks
many phenomena that the diffusive MC enjoys in three
dimensions. In particular, transmitted IMs are doomed to
hit the receiver, which is very different from the 3-D case,
where there is always the probability that no molecules will
reach the receiver. The extent of this simplification reveals
itself in the assumption that the transmitter releases a
single molecule per symbol, which, owing to the drift in
the channel and the absorbing nature of receiver, is always
detected.

MC-adapted version of the classical Hamming codes
was introduced in [218], where the traditional Hamming
distance metric on the codeword space, given by the
number of bit differences between two binary codewords,
is replaced by the so-called molecular coding distance
function (MoCo). In its essence, MoCo is defined in terms
of the negative logarithms of probabilities Pr({x → y})
of receiving codeword y when x was transmitted, and
the code aims at generating a codebook with maximal
minimum pairwise MoCo distance between constituent
codewords. MoCo distance is not a metric, as it is not
symmetric, and the triangular inequality is not verified
by the authors. This paper, too, considers a 1-D diffusive
channel with drift and an absorbing receiver; however,
in contrast to [216] and [217], it uses synchronized time-
slotted OOK modulation scheme with only a single type
of IM, and the receiver is additive with a threshold equal
to 1, i.e., it counts the number of hits in a period and
claims high logic reception with a single hit. In the case of
(4,2) block codes, the authors demonstrate that the code
generated using MoCo performs superior to the Hamming
code by carrying out an error rate analysis for both codes.
However, as shortcomings, MoCo depends on detection
probabilities that are sensitive to variations in channel
properties and are, in general, unknown to Tx and Rx.
Moreover, even if adaptive techniques may be envisioned,
calculation of the MoCo-based codebook (at Tx) and
the decoding region partition (at Rx) requires significant
computational resources at Tx and Rx, respectively, and
overhead communication would have to be considered for
the synchronized code updating.

Leeson and Higgins [219] propose the classical Ham-
ming codes to introduce error correction in MC, where they
consider a 3-D diffusive channel with time-slotted OOK
modulation scheme in a channel with finite memory. The
receiver is modeled as a nonabsorbing sphere that imme-
diately detects molecules that arrive at it, and reception
is decided upon the additive count of arriving molecules
during the transmission period. The Hamming codes are
error-correcting block codes with coding ratio k/n = (2m−
1)/(2m−m−1), where m is the number of parity check bits,
and [219] considers the Hamming codes for m = 3,4,5.
Their results show that the Hamming codes can deliver
coding gains up to ≈ 1.7 dB at a transmission distance
of 1 μm and for low BERs. Here, the coding gain is defined
as the gain the code introduces in a required number of

Vol. 107, No. 7, July 2019 | PROCEEDINGS OF THE IEEE 1321



Kuscu et al.: Transmitter and Receiver Architectures for MCs

IMs per transmission to achieve a given BER. At high BERs,
i.e., low quantities of transmitted IMs, extra ISI introduced
by parity bits overweighs error correction, and uncoded
transmission performs better. The authors also incorporate
an energy model for transmission, where energy is taken
to be proportional to the number of transmitted IMs. They
show that the energy required to transmit the extra parity
bits causes the coded transmission to be energy ineffi-
cient at small communication distances; however, coding
becomes more efficient at larger distances. Later on, over
the same channel model, a Hamming minimum energy
code (MEC) scheme was proposed in [220]. In a trade-
off of having larger codeword lengths against generating
codewords with lower average weights by using more 0-
bits, the authors trade between the rate and the energy
efficiency of communication. In subsequent works [221],
[222], again, over the same channel except with an absorb-
ing receiver in [222], self-orthogonal convolutional codes
(SOCCs) are proposed, and their performances against
Hamming MECs and uncoded transmissions are investi-
gated with respect to both BER and energy efficiency. Both
works conclude that in nanoscale, MC SOCCs have higher
coding gains, i.e., they are more energy efficient, compared
to the uncoded transmission and to the Hamming MECs
for the low BER (10−5–10−9) region. Moreover, SOCCs
are also reported to have shorter critical distances than
the Hamming MECs, where the critical distance is defined
as the distance, at which extra energy requirements of
employing coding are compensated by the coding gain.
Lu et al. [222] additionally explore the energy budget of
nano-to-macro and macro-to-nanomachine MCs and arrive
at the conclusion that in MC involving macromachines,
the critical distance of the codes decreases. Yet, in another
work [223], the Hamming codes are evaluated against
cyclic 2-D Euclidean geometry low-density parity-check
(EG-LDPC) and cyclic Reed–Muller codes by considering
the same channel model, as in [222]. Again, the com-
parison of codes is carried out for different MC scenar-
ios involving nanomachines and macromachines, and it
reveals, in the case of nano-to-nanomachine MC, that
in the BER region 10−3–10−6, the Hamming codes with
m = 4 are superior, and at lower BER regions, LDPC codes
with s = 2 exhibit the lowest energy cost. Here, s ≥ 2

is the density parameter in LDPC codes, where coding
density increases to 1 monotonically as s → ∞. Moreover,
in macro-to-nano and nano-to-macro MCs, the results indi-
cate that LDPC codes with s = 2 and s = 3 are the best
options, respectively.

The performance of convolutional coding techniques in
diffusive MC systems has been investigated by utilizing
PAM with M = 1, 2, 4 pulse amplitude levels for varying
key factors, such as transmission rate and communication
range (0.8 μm–1 mm) [224]. The findings indicate that
while convolutional coding with high transmission rate
and M = 1, i.e., OOK modulation, does outperform the
uncoded transmission in short- and medium-range com-
munications, no coding does better than convolutional

codes in the long-range MC. Furthermore, an increase in
the number of pulse amplitude levels causes deterioration
in achieved BERs, which is attributed to increased ISI,
implying that OOK modulation is better suited to MC than
PAM.

All the aforementioned works apply various channel
coding techniques for error correction in MC; however,
they do not provide any details into mechanisms of imple-
mentation of these codes from the device architecture
perspective. Marcone et al. [225] devised a molecular
single parity check (SPC) encoder with OOK modula-
tion for an MC design based on genetically engineered
bacteria that are assumed to network with each other
via signaling molecules, e.g., N-acyl homoserine lactones
(AHLs). The implementation of joint encoder–modulator
module is achieved via the design of genetic circuits
that regulate gene expression levels, and the transmission
materializes from ensuing biomolecule concentrations dic-
tated by biochemical reactions. Developed SPC encoder,
which appends to 2-bit information a parity check bit via
biological XOR gate based on designed genetic circuits,
provides an error detection mechanism, however with no
correction. Still, the introduced design serves as a basis
for genetic circuit-based designs of more complex block
codes with error correction capabilities. In this paper,
there is no evaluation of the proposed coding scheme,
as this paper considers only the transmitter side of MC.
A year later, Marcone et al. [131] extended their work in
[225] by introducing the biological analog decoder circuit,
which computes a posteriori log-likelihood ratio (LLR) of
transmitted bits from observed transmitter concentrations.
LLR is defined as the gain of the probability of detection
over nondetection in decibel. This enabled them to analyze
the whole end-to-end MC over a diffusive channel. Via sim-
ulations, they manage to verify the intended operation of
designed modulated SPC encoder and the analog decoder
and observe network performance close to an electrical
network operating in high noise.

V. D E T E C T I O N T E C H N I Q U E S F O R M C

Detection is one of the fundamental aspects of com-
munications having a tremendous impact over the overall
communication performance. The detection of MC signals
is particularly interesting due to the peculiarities of the MC
channel and communicating nanomachines, which impose
severe constraints on the design of detection methods. For
example, the limited energy budget and computational
capabilities of nanomachines due to their physical design
restrict the complexity of the methods. The memory of
the diffusion channel causes severe ISI and leads to time-
varying channel characteristics with very short coherence
time. The stochastic nature of the Brownian motion and
sampling of discrete message carriers bring about different
types of noise, e.g., counting noise and receptor binding
noise. The physiological conditions, in which most of the
nanonetwork applications are envisioned to operate, imply
the abundance of molecules with similar characteristics

1322 PROCEEDINGS OF THE IEEE | Vol. 107, No. 7, July 2019



Kuscu et al.: Transmitter and Receiver Architectures for MCs

Fig. 10. Fundamental aspects of MC detection investigated in this paper.

that can lead to strong molecular interference. These chal-
lenges have been addressed in MC to different extents.
In this section, we provide an overview of the state-of-the-
art MC detection approaches, along with a discussion on
their performances and weaknesses.

We classify the existing approaches according to the
considered channel and received signal models, which
reflects the envisioned device architectures that impose
different constraints or allow different simplifications over
the problem. Accordingly, we divide the detection meth-
ods into two main categories: MC detection with passive
and absorbing receivers and MC detection with reactive
receivers, as shown in Fig. 10.

A. MC Detection With Passive and
Absorbing Receivers

The nonlinearities and complexity of the MC system
often lead researchers to use simplifying assumptions
to develop detection methods and analyze their perfor-
mances. To this end, the intricate relationship between
molecular propagation and sampling processes is often
neglected.

Passive receiver (PA) concept is the most widely used
simplifying assumption in the MC literature, as it takes the
physical sampling process out of the equation, such that
researchers can focus only on the transport of molecular
messages to the receiver location. Accordingly, the passive
receiver is often assumed to be a spherical entity, whose
membrane is transparent to all kind of molecules, and it
is a perfect observer of the number of molecules within its
spherical reception space, as shown in Fig. 11 [226]. In the
passive receiver approximation, the receiver has no impact
on the propagation of molecules in the channel. Passive
receivers can also be considered, as if they include ligand
receptors that are homogeneously distributed within the
reception space with very high concentration and infinitely
high rate of binding with ligands, such that every single
molecule in the reception space is effectively bound to a
receptor at the time of sampling.

Another modeling approach, i.e., absorbing receiver
(AB) concept [227], considers receiver as a hypothetical
entity, often spherical, which absorbs and degrades every
single molecule that hits its surface, as demonstrated
in Fig. 11. This approach improves the assumption of
passive receiver one step further toward a more real-
istic scenario, including a physical interaction between
the receiver and the channel. In contrast to the passive
receiver, the absorbing receiver can be considered to have
receptors located over the surface. For a perfect absorbing
receiver, this means a very high concentration of receptors
with infinitely high absorption rate, such that every mole-
cule that hits the surface is bound and consumed instantly.

Physical correspondence of both models is highly ques-
tionable. Nevertheless, they are widely utilized in the liter-
ature, as they provide upper performance limits. However,
ignoring the receptor–ligand reactions, which often leads
to further intricacies, e.g., receptor saturation, stands as a
major drawback of these approaches.

1) Received Signal Models: When constructing the
received signal models for the diffusion-based MC,
the transmitter (Tx) geometry is usually neglected, assum-
ing that the Tx is a point source that does not occupy
any space. This assumption is deemed valid when the
distance between Tx and Rx is considerably larger than
the physical sizes of the devices. Throughout this section,
we will mostly focus on the OOK modulation, where the Tx
performs an impulsive release of a number of molecules to
transmit bit-1 and does not send any molecule to transmit
bit-0. This is the most widely used modulation scheme in
MC detection studies, as it simplifies the problem while
capturing the properties of the MC channel. However,
we will also briefly review the detection schemes corre-
sponding to other modulation methods, e.g., timing-based
modulation and MoSK, throughout this section.

Molecular propagation in the channel is usually assumed
to be only through free diffusion or through the combina-
tion of diffusion and uniform flow (or drift). In both cases,
the channel geometry is often neglected and assumed to

Vol. 107, No. 7, July 2019 | PROCEEDINGS OF THE IEEE 1323



Kuscu et al.: Transmitter and Receiver Architectures for MCs

Fig. 11. Hypothetical MC-Rx models used for developing detection

methods.

be unbounded, and molecules are assumed to propagate
independently of each other. In some studies addressing
passive receivers, researchers consider the existence of
enzymes in the channel, which reduces the impact of
the ISI by degrading the residual messenger molecules
through the first-order reaction [228]. For a 3-D free-
diffusion channel with uniform flow in the presence of
degrading enzymes, the number of molecules observed in
the spherical reception space of a passive receiver follows
nonstationary Poisson process [10], [229], that is

NRX|PA(t) ∼ Poisson (λRX(t)) (1)

where the time-varying mean of this process λRX(t) can be
given by

λRX(t) = λnoise + Q

� t
Ts

+1��
j=1

s[i]Pobs(t − (j − i)Ts). (2)

The mean depends on the number of transmitted mole-
cules Q to represent bit-1, the symbols transmitted in the
current symbol interval as well as in the previous symbol
intervals, i.e., s[i], and the length of a symbol interval
Ts. Most MC studies include an additive stationary noise
in their models, representing the interfering molecules
available in the channel as a result of an independent
process in the application environment. These molecules
are assumed to be of the same kind with the messenger

molecules, and their number is represented by a Pois-
son process and captured by λnoise. Channel response is
integrated into the model through the function Pobs(t),
which is the probability of a molecule transmitted at
time t = 0 to be within the sampling space at time t.
When Tx–Rx distance is considerably large, ligands are
typically assumed to be uniformly distributed within the
reception space. As a result, the channel response can be
written as

Pobs(t) =
VRX

(4πDt)3/2
exp

�
−kCEt − |�reff|2

4Dt

�
(3)

where VRX = (4/3)πd3
RX is the volume of the spherical

receiver with radius dRX, D is the diffusion coefficient, CE

is the uniform concentration of the degrading enzymes in
the channel, k is the rate of enzymatic reaction, and �reff

is the effective Tx–Rx distance vector, which captures the
effect of uniform flow [229]. Assuming that Tx and Rx are
located at �rTX = (0, 0, 0) and �rRX = (x0, 0, 0), respectively,
and the flow velocity is given by vx, vy , vz in 3-D Cartesian
coordinates, the magnitude of the effective distance vector
can be written as follows:

|�reff| =
�

(x0 − vxt)2 + (vy)+(vz)2. (4)

For an absorbing receiver, the received signal is usu-
ally taken as the number of molecules absorbed by the
Rx within a time interval [227]. For a diffusion channel
without flow, the probability density for a molecule emitted
at t = 0 to be absorbed by a perfectly absorbing receiver of
radius rr and located at a distance r from the Tx at time t

is given by

fhit(t) =
rr

r

1√
4πDt

r − rr

t
exp

�
− (r − rr)

2

4Dt

�
(5)

and the cumulative distribution function (CDF) is given
by

Fhit(t) =

� t

0

fhit(t
′)dt′ =

rr

r
erfc

�
r − rr√

4Dt

�
(6)

where erfc is the complementary error function [227]. The
CDF can be used to calculate the probability of a molecule
transmitted at time t = 0 to be absorbed within the kth
signaling interval, that is

Pk = Fhit(kTs) − Fhit([k − 1]Ts). (7)

When considering multiple independent molecules emitted
at the same time, the number of molecules absorbed at the
kth interval becomes the Bernoulli random variable with
the success probability of Pk. Assuming that the success
probability is low enough, the Gaussian approximation of

1324 PROCEEDINGS OF THE IEEE | Vol. 107, No. 7, July 2019



Kuscu et al.: Transmitter and Receiver Architectures for MCs

Table 4 Comparison Matrix for MC Detectors With Passive and Absorbing Receivers

the Bernoulli random process can be used to write

NRX|AB[i] ∼ N �
μ[i], σ2[i]

	
(8)

where its signal-dependent mean and variance can be
written as a function of current and previously transmitted
bits s[i], that is

μ[i] = Q
k�
i

Pks[i − k + 1] (9)

σ2[i] = σ2
noise + Q

k�
i

Pk(1 − Pk)s[i − k + 1]. (10)

Note that as in the case of passive receiver, the received
signal model includes the contribution of a stationary noise
through its variance σ2

noise. Unfortunately, in the litera-
ture, there is no analytical model for absorbing receivers
in diffusion-based MC channels with uniform flow and
degrading enzymes.

2) Detection Methods: Detection methods for MC,
in general, can be divided into two main categories
depending on the method of concentration measurement:
sampling- and energy-based detections. Passive receivers
are usually assumed to perform sampling-based detection,
which is based on sampling the instantaneous number of
molecules inside the reception space at a specific sampling

time [239]. Absorbing receivers, on the other hand, are
typically assumed to utilize energy-based detection, which
uses the total number of molecules absorbed by the
receiver during a prespecified time interval, that is usu-
ally the symbol interval [238]. In some studies, passive
receivers are also considered to perform energy-based
detection through taking multiple independent samples
of a number of molecules inside the reception space at
different time instants during a single-symbol interval and
passing them through a linear filter that outputs their
weighted sum as the energy of the received molecular
signal [229], [232].

As in conventional wireless communications, detection
can be done on symbol-by-symbol (SbS) or sequential
basis. The SbS detection tends to be more practical in terms
of complexity, whereas the sequence detectors require the
receiver to have a memory to store the previously decoded
symbols. Due to the MC channel memory causing a con-
siderable amount of ISI for high data rate communication,
the sequence detectors are more frequently studied in the
literature.

Next, we review the existing MC detection techniques
developed for passive and absorbing receivers by cate-
gorizing them into different areas depending on their
most salient characteristics. A comparison matrix for these
methods can also be seen in Table 4.

a) Symbol-by-symbol detection: SbS MC detectors in
the literature are usually proposed for very low-rate com-
munication scenarios, where the ISI can be neglected,

Vol. 107, No. 7, July 2019 | PROCEEDINGS OF THE IEEE 1325



Kuscu et al.: Transmitter and Receiver Architectures for MCs

asymptotically included into the received signal model
with a stationary mean and variance, or approximated by
the weighted sum of ISI contributions of a few previously
transmitted symbols. In [230], a one-shot detector is pro-
posed based on the asymptotic approximation of the ISI,
assuming that the sum of decreasing ISI contributions of
the previously transmitted symbols can be represented by a
Gaussian distribution through central limit theorem (CLT)
based on Lindeberg’s condition. A fixed-threshold detector
is proposed, maximizing the mutual information between
transmitted and decoded symbols. Similarly, in [229]
and [232], a matched filter in the form of a weighted
sum detector is proposed using a different asymptotic ISI
approximation, as, though, it results from a continuously
emitting source leading to a stationary Poisson distribution
of interference molecules inside the reception space. In this
scheme, a passive receiver performs energy-based detec-
tion taking multiple samples at equally spaced sampling
times during a single-symbol transmission, and the weights
of the samples are adjusted according to the number of
molecules expected at the corresponding sampling times.
This matched filter is proven to be optimal in the sense
that it maximizes SNR at the receiver. However, the opti-
mal threshold of this detector does not lend itself to a
closed-form expression, and thus, it should be numerically
obtained through resource-intensive search algorithms.
Similarly, in [231], considering also the external sources
of interference, another linear matched filter is designed,
maximizing the expected signal-to-interference-plus-noise
ratio (SINR) for SbS detection, and shown to outperform
previous schemes, especially when the ISI is severe. There
are also adaptive-threshold-based SbS detection methods
relying on receivers with a memory of varying length,
taking into account only the ISI contribution of a finite
number of previously transmitted symbols [233]–[236],
[238], [239]. In these schemes, the adaptive threshold is
updated for each symbol interval using the ISI estimation
based on the previously decoded symbols. SbS detection
is also considered in [243] and [245], which will be
discussed in the following in the context of noncoherent
and asynchronous detection.

b) Sequence detection and ISI mitigation: Optimal
sequence detection methods based on maximum a posteri-
ori (MAP) and maximum likelihood (ML) criteria are pro-
posed in [10] for MC with passive receivers. Even though
the complexity of the sequence detectors is reduced by
applying the Viterbi algorithm, it still grows exponentially
with increasing channel memory length. To reduce the
complexity further, a suboptimal linear equalizer based
on the minimum mean-square error (MMSE) criterion
is proposed. To improve the performance of the sub-
optimal detection, a nonlinear equalizer, i.e., decision-
feedback equalizer (DFE), is also proposed in the same
study. DFE is shown to outperform linear equalizers with
significantly less complexity than optimal ML and MAP
sequence detection methods. Similarly, a near-optimal
ML sequence detector employing the Viterbi algorithm is

proposed in [230]. Another optimal ML sequence detec-
tor is proposed in [232] for MC with uniform flow and
enzymes that degrade IMs.

In addition to the sequence detection methods and
equalizers, there are other approaches proposed to over-
come the effects of the ISI on detection. For example,
Akdeniz et al. [250] proposed to shift the sampling
time by increasing the reception delay to reduce the
effect of ISI. In [240], a derivative-based signal detection
method is proposed to enable high data rate transmis-
sion. The method is based on detecting the incoming
messages relying on the derivative of the channel impulse
response (CIR).

c) Noncoherent detection: Most of the MC detection
methods require the knowledge of the instantaneous CSI
in terms of CIR. However, CIR in MC, especially in physi-
ologically relevant conditions, tends to change frequently,
rendering the detection methods relying on the exact CIR
knowledge useless. Estimating the instantaneous CIR is
difficult and requires high computational power. To over-
come this problem, researchers propose low-complexity
noncoherent detection techniques. For example, Damrath
and Hoeher [236] develop a simple detection method
for absorbing receivers, which does not require chan-
nel knowledge. In this scheme, the receiver performs
a threshold-based detection by comparing the number
of absorbed molecules in the current interval to that
of the previous symbol interval. The adaptive threshold
is updated in every step of detection with the number
of molecules absorbed. However, this method performs
poorly when a sequence of consecutive bit-1s arrives.
Similarly, in [241], the difference of the accumulated con-
centration between two adjacent time intervals is exploited
for noncoherent detection. In [242], the local convexity of
the diffusion-based channel response is exploited to detect
MC signals in a noncoherent manner. A convexity metric
is defined as the test statistics, and the corresponding
threshold is derived. There are also methods requiring only
the statistical CSI rather than the instantaneous CSI [243].
In addition, constant-composition codes are proposed to
enable ML detection without statistical or instantaneous
CSI and shown to outperform uncoded transmission with
optimal coherent and noncoherent detection when the ISI
is neglected [244].

d) Asynchronous detection: The synchronization
between the communicating devices is another major
challenge. However, in the previously discussed studies,
synchronization is assumed to be perfect. To overcome
this limitation, an asynchronous peak detection method is
developed in [245] for the demodulation of MC signals.
Two variants have been proposed. The first method
is based on measuring the largest observation within
a sampling interval. This SbS detection method is of
moderate complexity and nonadaptive, comparing the
maximum observation to a fixed threshold. The second
method is adaptive and equipped with decision feedback
to remove the ISI contribution. In this scheme, the receiver

1326 PROCEEDINGS OF THE IEEE | Vol. 107, No. 7, July 2019



Kuscu et al.: Transmitter and Receiver Architectures for MCs

takes multiple samples per bit and adjusts the threshold
for each observation based on the expected ISI.

e) Detection for mobile MC: Majority of MC studies
assume that the positions of Tx and Rx are static during
communication. The mobility problem of MC devices has
just recently started to attract researchers’ attention. For
example, MC between a static transmitter and a mobile
receiver is considered in [246], where the authors pro-
pose to reconstruct the CIR in each symbol interval using
the time-varying transmitter–receiver distance estimated
based on the peak value of the sampled concentration.
Two adaptive schemes, i.e., concentration-based adaptive
threshold-detection and peak-time-based adaptive detec-
tion, are developed based on the reconstructed CIR.
In [247], different mobility cases, including mobile TX
and RX, mobile TX and fixed RX, and mobile RX and
fixed TX, are considered to develop a stochastic channel
model for diffusive mobile MC systems. The authors derive
analytical expressions for the mean, pdf, and autocorre-
lation function (ACF) of the time-varying CIR through an
approximation of the CIR with a log-normal distribution.
Based on this approximation, a simple model for outdated
CSI is derived, and the detection performance of a single-
sample threshold detector relying on the outdated CSI is
evaluated.

f) Other detection techniques: MC detection problem is
also addressed for MoSK modulation. In [237], an optimal
ML sequence detector employing the Viterbi algorithm is
proposed, assuming that a passive receiver can indepen-
dently observe MC signals carried by different types of
molecules. This assumption greatly simplifies the problem
and enables the application of detection methods devel-
oped for CSK-modulated MC signals for MoSK signals as
well.

Diffusion-based molecular timing (DBMT) channels are
also addressed from detection theoretical perspective.
DBMT channels without flow are accompanied by a Levy
distributed additive noise having a heavy algebraic tail in
contrast to the exponential tail of the inverse Gaussian
distribution, which DBMT channel with flow follows [249].
In [248], an optimal ML detector is derived for DBMT
channels without flow; however, the complexity of the
detector is shown to have exponential computational com-
plexity. Therefore, they propose suboptimal yet practical
SbS and sequence detectors based on the random time
of arrivals of the simultaneously released IMs and show
that the performance of the sequence detector is close
to the one of the computationally expensive optimal ML
detectors.

In DBMT channels without flow, linear filtering at the
receiver results in a dispersion larger or equal to the dis-
persion of the original, i.e., unfiltered, sample, rendering
the performance of releasing multiple particles worse than
releasing a single particle. Based on this finding, Murin
et al. [249] developed a low-complexity detector, which
is based on the first arrival (FA) time of the simultane-
ously released particles by the TX. The method is based

on the observation that the probability density of the FA
gets concentrated around the transmission time when the
number of released molecules M increases. Neglecting ISI,
it is shown in the same paper that the proposed FA-based
detector performs very close to the optimal ML detectors
for small values of M . However, the ML detection still
performs significantly better than the FA for high values
of M . The detection based on the order statistics has been
extended in the same authors’ later work [208], where
they consider also the detection based on the last arrival
(LA) time. Defining a system diversity gain as the asymp-
totic exponential decrease rate of error probability with the
increased number of released particles, they showed that
the diversity gain of the LA detector approaches to that of
computationally expensive ML detector.

B. MC Detection With Reactive Receivers

This type of receiver samples the molecular concen-
tration of incoming messages through a set of reactions
it performs via specialized receptor proteins or enzymes,
as shown in Fig. 11. The reactive receiver approach is
more realistic in the sense that natural cells, e.g., bacteria
and neurons, sense MC signals through their receptors on
the cell membrane, and many types of artificial biosensors,
e.g., bioFETs, are functionalized with biological receptors
for higher selectivity. Since synthetic biology, focusing on
using and extending natural cell functionalities, and artifi-
cial biosensing are the two phenomena that are considered
for the practical implementation of MC-Rxs, studying MC
detection with reactive receivers has more physical corre-
spondence.

Diffusion-based MC systems with reactive receivers,
in most cases, can be considered as reaction–diffusion (RD)
systems with finite reaction rates. Although RD systems,
which are typically highly nonlinear, have been studied in
the literature for a long time, they do not usually lend
themselves to analytical solutions, especially when the
spatiotemporal dynamics and correlations are not negligi-
ble. To be able to devise detection methods and evaluate
their performance in the MC framework, researchers have
come up with different modeling approaches, which will
be reviewed in the following. For the sake of brevity,
we focus our review on detection with receivers equipped
with ligand receptors, which have only one binding
site.

The ligand–receptor binding reaction for a single recep-
tor exposed to time-varying ligand concentration cL(t) can
be schematically demonstrated as follows:

U
cL(t)k+−−−−−⇀↽−−−−−

k−
B (11)

where k+ and k− are the ligand–receptor binding and
unbinding rates, respectively, and U and B denote the
unbound and bound states of the receptor, respectively.
When there are NR receptors, assuming that all of them

Vol. 107, No. 7, July 2019 | PROCEEDINGS OF THE IEEE 1327



Kuscu et al.: Transmitter and Receiver Architectures for MCs

are exposed to the same concentration of ligands, reaction
rate equation (RRE) for the number of bound receptors can
be written as follows [183]:

dnB(t)

dt
= k+cL(t) (NR − nB(t)) − k−nB(t). (12)

As is clear, while the binding reaction is second order
depending on the concentrations of both ligands and avail-
able receptors, unbinding reaction is first order and only
depends on the number of bound receptors.

Most of the time, the bandwidth of MC signals can
be assumed to be low enough to drive the binding reac-
tion to near equilibrium and allow applying quasi-steady-
state assumption for the overall system. In this case,
time-varying concentration cL(t) can be treated constant,
i.e., cL(t) = cL, and dnB(t)/dt = 0, which results in
the following expression for the mean number of bound
receptors:

E[nB ] =
cL

cL + KD
NR (13)

where KD = k−/k+ is the dissociation constant, which is a
measure of affinity between the specific type of ligand and
receptor. Even at equilibrium, the receptors randomly fluc-
tuate between the bound and unbound states. The number
of bound receptors nB at equilibrium is a Binomial random
variable with success probability pB = cL/(cL + KD), and
its variance can be given accordingly by

Var[nB ] = pB(1 − pB)NR. (14)

More insight can be gained by examining the
continuous history of binding and unbinding events
over receptors. The likelihood of observing a series
of n binding–unbinding events at equilibrium can be
given by

p({τB, τU}n)

=
1

Z
e
−

n

j=1τU
j (
�M

i=1 k+
i

ci)
n�

j=1

M�
i=1

k+
i cik

−
i e−k−

i
τB

j (15)

where Z is the normalization factor, τU
j and τB

j are the
jth unbound and bound time intervals, respectively, ci, k+

i ,
and k−

i are the concentration, binding rate, and unbinding
rate of the ith type of ligand, respectively, and M is the
number of ligand types present in the channel [251],
[252]. Note that the likelihood is equally valid for the cases
of single receptor and multiple receptors, as long as the
collected n samples of unbound and bound time intervals
are independent. These observable characteristics of the
ligand–receptor binding reactions have been exploited to
infer the incoming messages to different extents, as will be
reviewed in the following.

1) Received Signal Models: The nonlinearities arising
from the interaction of time-varying MC signals with recep-
tors have led to different approaches for modeling MC
systems with reactive receivers compromising on different
aspects to develop detection techniques and make the
performance analyses tractable. A brief review of these
modeling approaches is provided as follows.

a) Reaction–diffusion models with time-varying input:
One of the first attempts to model the ligand–receptor
binding reactions from an MC theoretical perspective is
provided in [183], where the authors develop a noise
model for the fluctuations in the number of bound recep-
tors of a receiver exposed to time-varying ligand concentra-
tions as MC signals. The model is based on the assumption
of a spherical receiver, in which ligand receptors and
information-carrying ligands are homogeneously distrib-
uted. For an analytically tractable analysis, the concentra-
tion of incoming ligands is assumed to be constant between
two sampling times, i.e., during a sampling interval, and
the ligand–receptor binding reaction is assumed to be at
equilibrium at the beginning of each sampling interval.
In light of these assumptions, the authors obtained the
time-varying variance and the mean of the number of
bound receptors, which are valid only for the correspond-
ing sampling interval. A more general approach without
the equilibrium assumption to obtain the mean num-
ber of bound receptors with time-varying input signals,
i.e., ligand concentration, is contributed by [253] and
[254] through solving the system of differential equations
governing the overall diffusion–reaction MC system. The
authors of both studies consider a spherical receiver with
ligand receptors on its surface and a point transmitter,
which can be anywhere on a virtual sphere centered at the
same point as the receiver but larger than that to obtain a
spherical symmetry to simplify the overall problem. As a
result, the transmitter location cannot be exactly speci-
fied in the problem. Deng et al. [254] considered that the
spherical receiver is capable of binding ligands at any point
on its surface, which is exactly equal to the assumption of
an infinite number of receptors. On the other hand, [253]
considers a finite number of receptors uniformly distrib-
uted on the receiver surface and addresses this challenge
through boundary homogenization. However, boundary
homogenization for a finite number of receptors does not
take into account the negative feedback of the bound
receptors on the second-order binding reaction [see (12)],
and thus, the developed analytical model is not able to
capture the indirect effects of a finite number of receptors,
e.g., receptor saturation. This is clear from their analysis,
such that the discrepancy between the analytical model
and the particle-based simulation results is getting larger
with increasing ligand concentration.

b) Frequency-domain model: Another modeling
approach is provided in [255], where the authors,
assuming that the probability of a receptor to be in the
bound state is very low, take the number of available,

1328 PROCEEDINGS OF THE IEEE | Vol. 107, No. 7, July 2019



Kuscu et al.: Transmitter and Receiver Architectures for MCs

i.e., unbound, receptors equal to the total number of
receptors at all time points. The complete first-order
characteristics of the resulting RRE enables them to
carry out a frequency-domain analysis, through which
they show that the ligand–receptor binding reaction
manifests low-pass filter characteristics. However, this
approximate model is relevant only when the probability
of receptor–ligand binding is very low.

c) Discrete model based on reaction–diffusion master
equation: To capture the stochasticity of the RD MC,
another approach is introduced in [256], where the
authors develop a voxel-based model based on the RD
master equation (RDME), with the diffusion and reactions
at the receiver modeled as the Markov processes. The
3-D MC system is discretized and divided into equal-size
cubic voxels, in each of which molecules are assumed to
be uniformly distributed, and allowed to move only to the
neighboring voxels. In the voxel accommodating the Tx,
the molecules are generated according to a modulation
scheme, and the Rx voxel hosts the receptor molecules,
where the ligands diffusing into the Rx voxel can react
based on a law of mass action. The jump of a ligand
from one voxel to another is governed by a diffusion rate
parameter, which is a function of the voxel size and the
ligand diffusion coefficient. The number of ligands and
bound receptors are stored in a system state vector, which
is progressed with a given state transition rate vector
storing the reaction, diffusion, and molecule generation
rates. In the continuum limit, the model is able to provide
closed-form analytical expressions for the mean and the
variance of the number of bound receptors for small-scale
systems. However, for larger systems, with a high number
of voxels, the efficiency of the model is highly questionable.

d) Steady-state model: In addition to the above-
mentioned approaches considering time-varying signals,
some researchers prefer using the assumption of steady-
state ligand–receptor binding reaction with stationary
input signals at the time of sampling, based on fact that
the bandwidth of incoming MC signals is typically low
because the diffusion channel shows low-pass filter char-
acteristics and the reaction rates are generally higher than
the diffusion rate of molecules. This assumption enables
the separation of the overall system into two: a deter-
ministic microscale diffusion channel and the stochastic
ligand–receptor binding reaction at the interface between
the receiver and the channel. Accordingly, at the sam-
pling time, the ligand concentration around the receptors
assumes different constant values corresponding to dif-
ferent symbols. The only fluctuations are resulting from
the binding reaction, where the random number of bound
receptors follows the Binomial distribution, whose mean
and variance are given in (13) and (14), respectively. The
steady-state assumption is applied in [257], where the
authors derive RD channel capacity for different settings.

e) Convection–diffusion–reaction system model:
Microfluidic MC systems with reactive receivers are
studied by a few researchers. Kuscu and Akan [137]

developed a 1-D analytical model, assuming that the
propagation occurs through convection and diffusion,
and a reactive receiver, which is assumed to be a SiNW
bioFET receiver with ligand receptors on its surface,
is placed at the bottom of the channel. The interplay
among convection, diffusion, and reaction is taken into
account by defining a transport-modified reaction rate,
tailored for the hemicylindrical surface of the SiNW
bioFET receiver. However, the authors assume steady-
state conditions for the reaction, to be able to derive
a closed-form expression for the noise statistics. The
molecular-to-electrical transduction properties of the
bioFET are reflected to the output current of the receiver
through modeling the capacitive effects arising from
the liquid–semiconductor interface and the 1/f noise
resulting from the defects of the SiNW transducer channel.
Kuscu and Akan [258] considered a 2-D convectional RD
system that does not lend itself to closed-form analytical
expressions for the received signal. The authors develop
a heuristic model using a two-compartmental modeling
approach, which divides the channel into compartments,
in each of which either transport or reaction occurs, and
derive an analytical expression for the time course of
the number of bound receptors over a planar receiver
surface placed at the bottom of the channel. The model
well captures the nonlinearities, such as Taylor–Aris
dispersion in the channel, depletion region above the
receiver surface, and saturation effects resulting from a
finite number of receptors, as validated through finite
element simulations of the system in COMSOL. However,
the model assumes that the channel and the receiver are
empty at the beginning of the transmission and, therefore,
does not allow an ISI analysis.

2) Detection Methods: The literature on detection meth-
ods for MC with reactive receivers is relatively scarce,
and the reason can be attributed partly to the lack of
analytical models that can capture the nonlinear ligand–
receptor binding reaction kinetics and resulting noise and
ISI. Nevertheless, the existing methods can be divided into
three categories depending on the type of assumptions
made and considered receiver architectures.

a) Detection based on instantaneous receptor states:
The first detection approach is based on sampling the
instantaneous number of bound receptors at a prespecified
time, as shown in Fig. 12, and comparing it to a threshold.
Deng et al. [254] studied a threshold-based detection
for OOK-modulated ligand concentrations using the dif-
ference between the number of bound molecules at the
start and the end of a bit interval. In [255], converting
the ligand–receptor binding reaction to a completely first-
order reaction with the assumption that all of the receptors
are always available for binding, the authors manage
to transform the problem into the frequency domain.
For the modulation, they consider MoSK with different
receptors corresponding to different ligands; therefore,
the problem basically reduces to a detection problem of the

Vol. 107, No. 7, July 2019 | PROCEEDINGS OF THE IEEE 1329



Kuscu et al.: Transmitter and Receiver Architectures for MCs

Fig. 12. Different methods for sampling receptor states in

reactive MC-Rxs.

concentration-encoded signals for each ligand–receptor
pair. To reduce the amount of noise, they propose to apply
a whitening filter to the sensed signal in the form of a num-
ber of bound receptors and then utilize the same detection
technique they proposed in [237]. An energy-based detec-
tion scheme is proposed in [259], where the test statistics
is the total number of binding events that occur within
a symbol duration. They propose a variable threshold-
detection scheme with varying memory length. The arti-
cle also takes into account the ISI; however, the model
assumes that all the receptors are always available for
binding and completely neglects the unbinding of ligands
from the receptors, making the reaction irreversible. Kuscu
and Akan [252] study the saturation problem in MC-Rxs
with ligand receptors. As a part of their analysis, they
investigated the performance of an adaptive threshold
ML detection using the instantaneous number of bound
receptors. The effect of the receiver memory that stores
the previously decoded bits is also investigated, and their
analysis reveals that the performance of MC detection
based on a number of bound receptors severely decreases
when the receptors get saturated, which occurs when they
are exposed to a high concentration of ligands as a result
of strong ISI. This is because near saturation, it becomes
harder for the receiver to discriminate between two levels
of a number of bound receptors corresponding to different
bits.

b) Detection based on a continuous history of recep-
tor states: The second detection approach is based on
exploiting the continuous history of binding and unbinding
events occurring at receptors or the independent sam-
ples of time intervals that the receptors stay bound and
unbound, as demonstrated in Fig. 12. As we see in the
likelihood function in (15), the unbound time intervals are
informative of the total ligand concentration, whereas the
bound time intervals are informative of the type of bound
molecules. This is exploited in [252], where the authors
tackle the saturation problem in reactive receivers. For
a receiver with ligand receptors and a memory storing

a finite number of previously decoded bits, they propose
to exploit the amount of time the individual receptors
stay unbound. Taking the ligand concentration station-
ary around the sampling time, and assuming steady-state
conditions for the ligand–receptor binding reaction, they
developed an ML detection scheme for OOK concentration-
encoded signals, which outperforms the detection based
on a number of bound receptors in the saturation case.
A simple intracellular reaction network to perform the
transduction of unbound time intervals into the concen-
tration of a certain type of molecules inside the cell is
also designed. In a similar manner, in [260] and [261],
using a voxel-based MC system model introduced in [256],
and by neglecting the ISI, the authors developed an opti-
mal MAP demodulator scheme based on the continuous
history of receptor binding events. Different from [252],
the authors assume the time-varying input signal. The
resulting demodulator is an analog filter that requires the
biochemical implementation of mathematical operations,
such as logarithm, multiplication, and integration. The
demodulator also needs to count the number of binding
events. In [260], they provide an extension of the demod-
ulator for the ISI case by incorporating decision feedback
and show the performance improvement with increasing
receiver memory.

One of the challenges of reactive receivers is their
selectivity toward the messenger molecules, which is not
perfect in practice. It is highly probable, especially in phys-
iologically relevant environments, that there are similar
ligands in the channel, which can also bind the same
receptors, even though their unbinding rate is higher than
that of the correct, i.e., messenger, ligands. This causes
a molecular interference, which impedes the detection
performance of the receiver, especially when the concen-
tration of interferer molecules is not known to the receiver.
Muzio et al. [262] evaluated the performance of the one-
shot ML detection schemes based on the number of bound
receptors and unbound time intervals in the presence of
interference from a similar ligand available in the environ-
ment, number of which in the reception space follows the
Poisson distribution, with a known mean. They proposed a
new ML detection scheme based on estimating the ratio of
messenger ligands to the inferring ligands using the bound
time intervals sampled from each receptor. It is shown that
the proposed method substantially outperforms the others,
especially when the concentration of interferer molecules
is very high. Note that the above-mentioned detection
techniques are built on MC models that neglect the receiver
geometry by treating it as a transparent receiver, inside
which ligands and receptors are homogeneously distrib-
uted, and they require the receiver to store an internal
model, i.e., CIR, of the RD channel.

c) Detection for biosensor-based MC receivers: The third
set of detection methods deals with the biosensor-based
receivers, where the binding events are transduced into
electrical signals. In bioFET-based receivers, the concen-
tration of bound charge-carrying ligands is converted into

1330 PROCEEDINGS OF THE IEEE | Vol. 107, No. 7, July 2019



Kuscu et al.: Transmitter and Receiver Architectures for MCs

electrical signals that are contaminated with additional
noise. It is not possible to observe individual receptors
states; therefore, the detection based on a continuous his-
tory of binding events is not applicable to these receivers.
Accounting for the 1/f noise and binding fluctuations at
steady-state conditions, Kuscu and Akan [137] developed
an optimal ML detection scheme for CSK in the absence
of ISI. Approximating the binding and 1/f noise with a
Gaussian distribution, they reduced the overall problem to
a fixed-threshold-detection problem and provided closed-
form analytical expressions for the optimal thresholds and
corresponding symbol error rates. The performance eval-
uation reveals that the 1/f noise, which is resulting from
the defects of the semiconductor FET channel, surpasses
the binding noise resulting from the fluctuations of the
receptor states, especially at low frequencies, and severely
degrades the detection performance.

VI. C H A L L E N G E S A N D F U T U R E
R E S E A R C H D I R E C T I O N S

After providing a comprehensive survey of existing stud-
ies on different aspects of MC, in this section, we discuss
the most important challenges toward the realization of
microscale/nanoscale MC setups and evaluation of their
ICT-based performance. In this direction, we highlight
both the required theoretical studies and the experimental
investigations to validate the theoretical models. The phys-
ical architecture of MC-Tx/Rx is one of the least studied
topics in the MC literature; however, it significantly affects
the accuracy of theoretical studies. Thus, we provide an
in-depth discussion on the required investigations of the
MC Tx/Rx architecture. The second shortcoming of the MC
literature is the use of nonrealistic assumptions in the
existing MC-Tx/Rx and channel models, which has led to
imprecise performance estimations not applicable to real
scenarios. Hence, we explain the requirements of realistic
modeling, such as taking into account the impact of the
stochastic molecule generation process, Tx/Rx geometry,
and stochastic noise dynamics. We also describe open
research problems in the developed modulation, coding,
and detection techniques for MC and emphasize important
factors that must be considered in future investigations.

A. Challenges for Physical Design and
Implementation of MC-Tx

Theoretical modeling of MC-Tx in the literature gener-
ally relies on unrealistic simplifying assumptions due to
the lack of experimental studies on microscale/nanoscale
implementation of MC-Tx. Therefore, MC-Tx is often
assumed as an ideal point source capable of perfectly trans-
mitting molecular messages, which completely neglects
important factors in the IM release process, such as the
stochasticity in the molecule generation process, the effect
of Tx geometry, and the channel feedback. As modeling
microscale/nanoscale MC-Tx without any experimental
data seems not possible, the requirement for empirical

transmitter models is the most pressing challenge for MC,
as end-to-end channel models are required to consider the
effects of both MC-Tx and MC-Rx. Based on the empirical
channel models, the communication theoretical investiga-
tion of MC, i.e., analysis of capacity, modulation, coding,
and detection, needs to be revisited, as communication
parameters show a strong dependence on the channel
model. Other challenges regarding the practical imple-
mentation of a microscale/nanoscale MC-Tx are listed as
follows.

1) Energy: Microscale/nanoscale MC-Tx and MC-Rx are
required to work as stand-alone devices with their
own energy sources. Since battery-powered devices
have limited lifetime, EH techniques are promising
to develop energy neutral devices, such that all
operations of the device can be powered via har-
vested energy from various sources, such as solar,
mechanical, and chemical. To design such systems,
one needs to first calculate the harvestable energy
budget of an MC-Tx and an MC-Rx, i.e., the amount
of energy harvestable from the surrounding based
on application scenarios and medium. Then, both
transmitter and receiver operations, e.g., modula-
tion, coding, and detection, are required to be devel-
oped, accordingly not to exceed the available energy
budget. Considering miniaturized MC-Tx and MC-
Rx, the harvestable energy can be quite limited so
that complex algorithms may not be feasible in a
realistic MC scenario.

2) Data Rate: MC is mostly promising in applications
without high data rate requirements, as MC suf-
fers from slow propagation channels. This prob-
lem can be tackled by encoding a large amount of
data in DNA/RNA strands, which is named NSK,
as discussed in Section IV-A. This way, the amount
of transmitted information can be increased up to
hundereds of megabytes per IM, such that MC can
achieve data rates on the order of megabits per sec-
ond. Although sequencing of DNA/RNA strands
can be performed via stand-alone devices utilizing
nanopores, there is yet any practical low-cost system
to write DNA sequences with a microscale/nanoscale
device.

3) Molecule Leakage: In the case of nanomaterial-
based MC-Txs, there are two significant design
challenges: molecule leakage and molecule reser-
voir/generation. Molecule leakage, while transmit-
ting low logic or no information, is unavoidable
for practical MC-Tx designs. The leaking molecules
increase ISI in the channel and also contribute to an
additional problem in molecule reservoir/generation
by lowering the molecule budget of MC-Tx. In order
to tackle this problem, we have proposed some
solutions based on the molecule wax as an IM con-
tainer and hybrid MC-Tx designs, where genetically
engineered bacteria can be utilized to replenish IM
sources. However, these solutions have not been

Vol. 107, No. 7, July 2019 | PROCEEDINGS OF THE IEEE 1331



Kuscu et al.: Transmitter and Receiver Architectures for MCs

implemented, and the feasibility of such systems as
MC-Tx still stands as an important open research
issue.

4) Biological Complexity: In the case of biological
MC-Tx architectures, the main difficulty in design
stems from the complexity of involved biological
elements. Understanding of molecular basis under-
lying cellular mechanisms, i.e., cellular biology, and
development of techniques to manipulate them, i.e.,
synthetic biology, are essential to unlock the reli-
able use of engineered biological entities for MC.
Moreover, as, in many applications, data propagated
in nanonetworks eventually need to be connected
to electronic devices, any biological network archi-
tecture must be interfaced with an electronic archi-
tecture. Thus, the development of MC architectures
within this space requires an extremely high inter-
disciplinary engagement between the fields of ICT,
biology, and synthetic biology, as well as materials
science and nanofabrication. Another issue, which,
again, is caused by the inherent complexity of bio-
logical organisms, is the fact that genetically engi-
neered cells are not the best survivors, and compli-
cations in their engineered metabolisms cause the
accelerated death of the cell.

B. Challenges for Physical Design and
Implementation of MC-Rx

Despite existing theoretical studies on the performance
of MC and few macroscale experimental setups, the lit-
erature lacks comprehensive investigation of the physical
design of microscale/nanoscale MC-Rx structures. This
leads to critical open challenges needed to be tackled for
the realization of the MC promising applications.

1) Ligand–Receptor Selection: As mentioned in
Section III-C, both genetic circuit-based and
artificial architectures use ligand–receptor reactions
to sense the concentration of target molecule
by the MC antenna. This calls for the careful
selection of appropriate ligand–receptor pairs for
the MC paradigm. In this regards, the binding and
dissociation, i.e., unbinding, rates of these reactions
are among the important parameters that must be
taken into consideration. The binding rate controls
sensitivity, selectivity, and the response time of
the device. While high dissociation rates reduce
the reusability of the device, very low values are
also not desired, as they decrease the sensitivity
of the receiver. The existing interferer molecules
in the application environment and their affinities
with the receptors are the next important design
parameter that must be studied to minimize the
background noise.

2) Realistic ICT-Based Modeling of Artificial Structures:
For artificial biosensor-based MC-Rxs, the litera-
ture is lacking analytical models that can capture

transient dynamics, as the available models devel-
oped from the sensor application perspective are
mostly based on the equilibrium assumption. How-
ever, in MC applications, since the concentration sig-
nals are time-varying, equilibrium models may not
be realistic. Therefore, devising MC detection meth-
ods for artificial receivers requires developing more
complex models that can also capture the stochastic
noise dynamics without steady-state assumption for
different types of biosensors based on nanomateri-
als, e.g., graphene bioFET. The models should also
include the effects of operating voltages for bioFETs,
receiver geometry, and gate configuration, and chan-
nel ionic concentration determining the strength of
the Debye screening. The models should be vali-
dated through wet-lab experiments, and microfluidic
platforms stand as a promising option for imple-
menting test beds for MC systems with artificial
MC-Rxs.

3) Biological Circuits Complexity: In the case of imple-
menting MC-Rxs with genetic circuits, the infor-
mation transmission is through molecules and
biochemical reactions, which results in nonlin-
ear input–output behaviors with system-evolution-
dependent stochastic effects. This makes the
analytically studying the performance metrics diffi-
cult if not impossible. Moreover, the existing noise
in genetic circuits must be comprehensively studied,
and methods to mitigate this noise must be derived,
as it significantly reduces the achievable mutual
information of the MC [135].

4) Bio–Cyber Interface: While MC expands the function-
ality of nanomachines by connecting them to each
other and making nanonetworks [263], the con-
nection of these nanonetworks to the cybernet-
works, i.e., making IoBNT, further extends the
applications of these nanomachines [1]. Continuous
health monitoring and bacterial sensor–actor net-
works inside the human body are two promising
examples of these applications. To this aim, imple-
mentation of microscale/nanoscale bio–cyber inter-
faces is required. The bio–cyber interface needs to
decode the molecular messages, process it, and send
the decoded information to a macroscale network
node through a wireless link. In bioFET-based MC-
Rxs, owing to the molecular-to-electrical transducer
unit, the electrical signal generated by the receiver
can be sent to the cybernetworks through EM wire-
less communications. However, the connection of
biological MC-Rxs to the cybernetworks remains as
an open issue.

Finally, it is worthwhile to mention that the physical
architecture of the device is mainly dictated by the appli-
cation requirements. As an example, biological MC-Rxs
are only promising and feasible for in vivo applications.
On the other hand, utilizing artificial structures for in vivo
applications necessitate a comprehensive investigation of

1332 PROCEEDINGS OF THE IEEE | Vol. 107, No. 7, July 2019



Kuscu et al.: Transmitter and Receiver Architectures for MCs

the receiver biocompatibility. Moreover, the physiological
conditions imply solutions with high ionic concentrations,
an abundance of interferers and contaminants, and the
existence of disruptive flows and fluctuating tempera-
ture, which may degrade the receiver’s performance in
several aspects. First, high ionic concentration creates a
strong screening effect, reducing the Debye length, thus
impedes the sensitivity of the receiver [187]. Moreover,
contaminants and disruptive flows may alter the binding
kinetics, impede the stability of the receptors, and, even,
separate them from the dielectric layer. These call for
the comprehensive investigation of these factors, their
impact on the performance of the device, and methods
to control their effects. To control the screening problem,
using highly charged ligands and very small size recep-
tors, such as aptamers, are theoretically efficient. How-
ever, the frequency-domain technique promises for much
more realistic solutions. Zheng et al. [264] revealed that
frequency-domain detection outperforms the conventional
time-domain technique in terms of sensitivity in highly
ionic solutions. An alternative solution to overcome the
Debye screening limitations in detection is proposed in
[265], where it is shown that applying a high-frequency
alternating current between the source and the drain
electrodes, instead of a dc current, weakens the double-
layer capacitance generated by the solution ions. They
show that the effective charges of the ligands become
inversely dependent on the Debye length instead of the
exponential dependence. As a result, the screening effect
on the bound ligands is significantly reduced, and the FET-
based biosensing becomes feasible even for physiological
conditions. Similar approaches are taken by others to over-
come the Debye screening with radio frequency operation
of graphene bioFETs, which are summarized in [266].

C. Challenges for Developing
MC Modulation Techniques

There are various modulation schemes that are pro-
posed for MC by utilizing concentration, molecule
type/order/ratio, and release timing. However, these stud-
ies mostly utilize simplified channel models based on MC-
Tx, which is an ideal point source, and MC-Rx, which is
capable of perfectly detecting multiple molecules selec-
tively. Therefore, the performance of the proposed schemes
under realistic conditions is still unknown, and this can
be tackled by implementing the proposed schemes in an
experimental MC setup. In addition, some of the mod-
ulation schemes require synchronization, which is hard
to achieve considering the error-prone diffusive channel
and low-complexity MC devices at microscale/nanoscale.
Furthermore, energy efficiency is another important chal-
lenge for MC-Tx designs as being powered via limited
energy harvested from the medium. Therefore, energy-
efficient and low-complexity modulation schemes for MC
without strict synchronization requirements still stand as a
significant research problem.

D. Challenges for Developing MC Channel
Coding Techniques

Even though there is some literature on MC channel
coding techniques, the field is still very new and open
for research. Some of the most pressing directions are
highlighted in the following.

1) Adaptive Coding: The propagation time through MC
channel and detection probabilities at the receiver
are affected by various environmental parameters,
such as temperature, diffusion speed, channel con-
tents, reaction rates, and distance between nodes,
which calls for adaptive channel coding techniques
for error compensation [267]. In this respect, none
of the coding schemes investigated even considers to
probe the channel for its characteristics in order to
adapt itself.

2) Irregular Signaling: All of the coding schemes
discussed have been evaluated under the regular
time-slotted transmission assumption, which, in any
realistic MC scenario, will never be the case. For
instance, fluctuations in the intersymbol duration
caused by irregular transmission would have a sig-
nificant effect on suffered ISI, the primary source of
BER in MC, and, therefore, needs to be considered
by channel codes.

3) Simplicity of Models: More channel codes for MC
need to be invented. The only works that do so
are [216]–[218], and they have very simple trans-
mission and channel models, i.e., in all works,
the transmitter releases only a single molecule per
transmission period, and the channel is 1-D.

E. Challenges for Developing
MC Detection Techniques

As reviewed in Section V, the MC detection prob-
lem has been widely addressed from several aspects for
different channel and receiver configurations. However,
the proposed solutions are still far from being feasible
for envisioned MC devices. The main reason behind this
discrepancy between the theoretical solutions and the real
practice, which is also revealed by the preliminary exper-
imental airborne MC studies performed with macroscale
off-the-shelf components [68], [268], is that there is no
microscale/nanoscale MC test bed that can be used as a
validation framework to optimize the devised methods.
In parallel to this general problem regarding MC tech-
nologies, major challenges for MC detection can be further
detailed as follows.

1) Synchronization: The majority of the proposed detec-
tion techniques assume a perfect synchronization
between the transmitter and the receiver. In fact,
synchronization is essential for the proper operation
of the proposed solutions. There are many studies
proposing different synchronization methods, e.g.,
using quorum sensing to globally synchronize the
actions of nanomachines in a nanonetwork [269],

Vol. 107, No. 7, July 2019 | PROCEEDINGS OF THE IEEE 1333



Kuscu et al.: Transmitter and Receiver Architectures for MCs

blind synchronization and clock synchronization
based on the ML estimation of the channel
delay [270], [271], and peak observation-time-
and threshold-trigger-based symbol synchronization
schemes employing a different type of molecule
for the purpose of synchronization [272]. However,
these methods are either too complex for the limited
capabilities for the nanomachines or they rely on
stable CSI, which is not the case for time-varying MC
channels. Moreover, the effect of these nonideal syn-
chronization techniques on the performance of the
proposed detection methods has not been revealed.
Another potential solution to the synchronization
problem could be to develop asynchronous detection
techniques that obviate the need for synchroniza-
tion. As reviewed in Section V-A, there are a few
promising solutions for asynchronous MC detection,
e.g., peak-detection and threshold-based detection
methods [245]. However, they are mostly built on
simplifying assumptions for receiver and channel
geometry and properties, which may result in unex-
pected performance in real applications.

2) Physical Properties of the Receiver: Although there
is a little physical correspondence for passive and
absorbing receivers, many of the detection schemes
are built on these assumptions, as they enable a
mathematically tractable analysis. As reviewed in
Section V-B, although they consider the effect of
receptor reactions, the initial studies on reactive
receivers also follow similar assumptions on the
device architecture and the geometry to simplify the
analyses. However, for practical systems, the phys-
ical properties of the realistic receiver architectures
and their impact on the molecular propagation in the
MC channel should be taken into consideration to
the most possible extent. Finite element simulations
on microfluidic MC channel with bioFET receivers
and macroscale MC experiments with alcohol sen-
sors clearly reveal the effect of the coupling between
the MC-Rx and the channel [258], [268]. The cou-
pling is highly nonlinear, and in most of the cases,
it is not analytically tractable; therefore, beyond the
available analytical tools, researchers may need to
focus on stochastic simulations and experiments to
validate the performance of the proposed detection
techniques in realistic scenarios.

3) Physical Properties of the Channel: Most of the MC
detection studies assume free diffusion, or diffusion
plus uniform flow, for molecular propagation in an
unbounded 3-D environment. However, in practice,
MC channel will be bounded with varying boundary
conditions and can include nonuniform and dis-
ruptive flows, obstacles, temperature fluctuations,
charged molecules affecting the diffusion coeffi-
cient, and particles leading to channel crowding.
Some of these aspects have recently started to be
addressed through channel modeling studies, e.g.,

subdiffusive MC channel due to molecular crowding
[273] and diffusion-based MC in multilayered bio-
logical media [274]. However, these simplified mod-
els are not able to provide enough insight into the
detection problem in realistic channels. The highly
time-varying properties of the MC channel have
also been addressed by researchers through channel
estimation techniques [275], [276] and noncoher-
ent detection techniques [241]–[243], which are
reviewed in Section V-A. These studies are built on
simplifying the assumptions on channel and receiver
architecture.

4) Reactive Receivers: Although reactive receiver con-
cept provides a more realistic approach to the detec-
tion problem, the research in this direction is still at
its infancy, and the developed received signal models
are not complex enough to reflect many intricacies.
For example, most of the previous studies assume
independent receptors being exposed to the same
ligand concentration; however, this is not always the
case, as the binding of one receptor can affect the
binding of neighboring ones [277], and receptors
can form cooperative clusters to control sensitivity
by exploiting spatial heterogeneity [278], [279].
The interplay between diffusion and reaction is also
often neglected in these studies by assuming that
the timescales of both processes are separated suf-
ficiently. However, in most practical cases, reaction
and diffusion rates are close to each other, and
reactions are correlated with the transport process,
which depends on the channel properties [182],
[280]. These problems are crucial to analyze the spa-
tiotemporal correlations among the receptors and
the coupling between the diffusion channel and the
reactive receiver. Moreover, except for a few recent
studies [131], [281], [282], the design of intracel-
lular reaction networks to implement the proposed
detectors is usually neglected. The additional noise
and delay stemming from these reaction networks
should be taken into account while evaluating the
performance of the overall detection. Furthermore,
a proper analysis of the tradeoff among energy,
detection accuracy, and detection speed is required
to develop an optimization framework for the detec-
tor design in reactive receivers.

5) Receiver Saturation: Another challenge associated
with reactive receivers is the saturation of recep-
tors in the case of strong ISI or external interfer-
ence, which can severely limit the receiver dynamic
range and hamper the ability of the receiver to
discriminate different signal levels. The saturation
problem has been recently addressed in [252]
through the steady-state assumption for the ligand–
receptor binding reactions; however, in general, it is
neglected to assume an infinite number of receptors
on the receiver. The problem can also be alleviated
by adaptive threshold-detection techniques and

1334 PROCEEDINGS OF THE IEEE | Vol. 107, No. 7, July 2019



Kuscu et al.: Transmitter and Receiver Architectures for MCs

adaptive transmission schemes as in detection with
passive/absorbing receivers.

6) Receiver Selectivity: Receiver selectivity is also a
major issue for MC detection, and it has just started
to be addressed from the MC perspective. The phys-
iologically relevant environments usually include
many similar types of molecules, and the receptor–
ligand coupling is not typically ideal, such that many
different types of ligands present in the channel can
bind the same receptors as the messenger ligands,
causing molecular interference. Even if there are
different types of receptor molecules for each ligand
type in an MC system with MoSK, the crosstalk
between receptors is unavoidable because of the
nonideal coupling between receptor and ligands.
Therefore, there is a need for selective detection
methods exploiting the properties of ligand–receptor
binding reaction. The amount of time a receptor
stays bound is informative of the ligand unbind-
ing rate, which is directly linked with the affin-
ity between the particular types of ligand–receptor
pairs. The bound time intervals are exploited in
[262] to detect the MC messages based on the
ML estimation of the ratio of correct ligands to
the interferers. However, this technique requires the
receiver to know the type and the probability distri-
bution of the interferer molecules, and it is devel-
oped only for one type of interferer. To implement
selective detection methods in engineered bacteria-
based MC devices, the kinetic proofreading and
adaptive sorting techniques implemented by reac-
tion networks of the T cells in the immune sys-
tem can be exploited [251], [283]. In addition,
MC spectrum sensing methods can be developed
based on the information inferred from the receptor
bound time intervals to apply cognitive radio tech-
niques in crowded MC nanonetworks, where many
nanomachines communicate using the same type of
molecules [284].

VII. C O N C L U S I O N

MC has attracted significant research attention over the
last decade. Although ICT aspects of MC, i.e., information
theoretical models of MC channels and modulation and
detection techniques for MC and system theoretical mod-
eling of MC applications, are well-studied, these research
efforts mostly rely on the unrealistic assumptions, isolating
the MC channel from the physical processes regarding
transmitting and receiving operations. The reason being
that although there are some proposals for MC-Tx/Rx
based on synthetic biology and nanomaterials, there is no
implementation of any artificial microscale/nanoscale MC
system to date. As a result, the feasibility and performance
of ICT techniques proposed for MC could not be validated.

In this paper, we provide a comprehensive survey on
the recent proposals for the physical design of MC-Tx/Rx
and the state of the art in the theoretical MC research,
covering modulation, coding, and detection techniques.
We first investigate the fundamental requirements for
the physical design of microscale/nanoscale MC-Tx/Rx in
terms of energy and molecule consumption and operat-
ing conditions. In light of these requirements, the state-
of-the-art design approaches as well as novel MC-Tx/Rx
architectures, i.e., artificial Tx/Rx designs enabled by
the nanomaterials, e.g., graphene, and biological Tx/Rx
designs enabled by synthetic biology, are covered. In addi-
tion, we highlight the opportunities and the challenges
regarding the implementation of Tx/Rx and corresponding
ICT techniques that are to be built on these devices.

The guidelines on the physical design of MC-Tx/Rxs
provided in this paper will help researchers to design
experimental MC setups and develop realistic channel
models, considering the transceiving processes. In this way,
the long-standing discrepancy between theory and practice
in MC can be overcome toward unleashing the huge soci-
etal and economic impact of MC by paving the way for
ICT-based early diagnosis and treatment of diseases, such
as IoBNT-based continuous health monitoring, smart drug
delivery, artificial organs, and lab-on-a-chip.

R E F E R E N C E S
[1] I. F. Akyildiz, M. Pierobon, S. Balasubramaniam,

and Y. Koucheryavy, “The Internet of bio-nano
things,” IEEE Commun. Mag., vol. 53, no. 3,
pp. 32–40, Mar. 2015.

[2] I. F. Akyildiz and J. M. Jornet, “The Internet of
nano-things,” IEEE Wireless Commun., vol. 17,
no. 6, pp. 58–63, Dec. 2010.

[3] O. B. Akan, H. Ramezani, T. Khan, N. A. Abbasi,
and M. Kuscu, “Fundamentals of molecular
information and communication science,” Proc.
IEEE, vol. 105, no. 2, pp. 306–318, Feb. 2017.

[4] L. Felicetti, M. Femminella, G. Reali, and P. Liò,
“Applications of molecular communications to
medicine: A survey,” Nano Commun. Netw., vol. 7,
pp. 27–45, Mar. 2016.

[5] B. Atakan, O. B. Akan, and S. Balasubramaniam,
“Body area nanonetworks with molecular
communications in nanomedicine,” IEEE Commun.
Mag., vol. 50, no. 1, pp. 28–34, Jan. 2012.

[6] N. Farsad, H. B. Yilmaz, A. Eckford, C.-B. Chae,
and W. Guo, “A comprehensive survey of recent
advancements in molecular communication,” IEEE
Commun. Surveys Tuts., vol. 18, no. 3,

pp. 1887–1919, 3rd Quart., 2016.
[7] S. Qiu, W. Guo, S. Wang, N. Farsad, and

A. Eckford, “A molecular communication link for
monitoring in confined environments,” in Proc.
IEEE Int. Conf. Commun. Workshops (ICC),
Jun. 2014, pp. 718–723.

[8] M. Pierobon and I. F. Akyildiz, “Capacity of a
diffusion-based molecular communication system
with channel memory and molecular noise,” IEEE
Trans. Inf. Theory, vol. 59, no. 2, pp. 942–954,
Feb. 2013.

[9] M. S. Kuran, H. B. Yilmaz, T. Tugcu, and
I. F. Akyildiz, “Modulation techniques for
communication via diffusion in nanonetworks,” in
Proc. IEEE Int. Conf. Commun. (ICC), Jun. 2011,
pp. 1–5.

[10] D. Kilinc and O. B. Akan, “Receiver design for
molecular communication,” IEEE J. Sel. Areas
Commun., vol. 31, no. 12, pp. 705–714,
Dec. 2013.

[11] Y. Chahibi, “Molecular communication for drug
delivery systems: A survey,” Nano Commun. Netw.,
vol. 11, pp. 90–102, Mar. 2017.

[12] D. Daksh, D. Rawtani, and Y. K. Agrawal, “Recent
developments in bio-nanoelectronics devices:
A review,” J. Bionanosci., vol. 10, no. 2, pp. 81–93,
2016.

[13] Q. H. Abbasi et al., “Nano-communication for
biomedical applications: A review on the
state-of-the-art from physical layers to novel
networking concepts,” IEEE Access, vol. 4,
pp. 3920–3935, 2016.

[14] R. B. Schoch, J. Han, and P. Renaud, “Transport
phenomena in nanofluidics,” Rev. Mod. Phys.,
vol. 80, no. 3, p. 839, 2008.

[15] I. F. Akyildiz, F. Brunetti, and C. Blázquez,
“Nanonetworks: A new communication
paradigm,” Comput. Netw., vol. 52, no. 12,
pp. 2260–2279, 2008.

[16] J.-T. Huang and C.-H. Lee, “On capacity bounds of
two-way diffusion channel with molecule
harvesting,” in Proc. IEEE Int. Conf.
Commun. (ICC), May 2017, pp. 1–6.

[17] H. G. Bafghi, A. Gohari, M. Mirmohseni, and
M. Nasiri-Kenari. (2018). “Diffusion based
molecular communication with limited molecule

Vol. 107, No. 7, July 2019 | PROCEEDINGS OF THE IEEE 1335



Kuscu et al.: Transmitter and Receiver Architectures for MCs

production rate.” [Online]. Available:
https://arxiv.org/abs/1802.08965

[18] B. D. Unluturk, A. O. Bicen, and I. F. Akyildiz,
“Genetically engineered bacteria-based
biotransceivers for molecular communication,”
IEEE Trans. Commun., vol. 63, no. 4,
pp. 1271–1281, Apr. 2015.

[19] A. B. Bilgin and O. B. Akan, “A fast algorithm for
analysis of molecular communication in artificial
synapse,” IEEE Trans. Nanobiosci., vol. 16, no. 6,
pp. 408–417, Sep. 2017.

[20] C. Pin̄ero-Lambea, D. Ruano-Gallego, and
L. Á. Fernández, “Engineered bacteria as
therapeutic agents,” Current Opinion Biotechnol.,
vol. 35, pp. 94–102, Dec. 2015.

[21] S. Haefner et al., “Chemically controlled
micro-pores and nano-filters for separation tasks
in 2D and 3D microfluidic systems,” RSC Adv.,
vol. 7, no. 78, pp. 49279–49289, 2017.

[22] C. Di Natale et al., “Molecularly endowed
hydrogel with an in silico-assisted screened
peptide for highly sensitive small molecule
harvesting,” Chem. Commun., vol. 54, no. 72,
pp. 10088–10091, 2018.

[23] E. Katz, Implantable Bioelectronics. Hoboken, NJ,
USA: Wiley, 2014.

[24] S. Sharifi, S. Behzadi, S. Laurent, M. L. Forrest,
P. Stroeve, and M. Mahmoudi, “Toxicity of
nanomaterials,” Chem. Soc. Rev., vol. 41, no. 6,
pp. 2323–2343, 2012.

[25] F. Asghari, M. Samiei, K. Adibkia, A. Akbarzadeh,
and S. Davaran, “Biodegradable and
biocompatible polymers for tissue engineering
application: A review,” Artif. Cells, Nanomed.,
Biotechnol., vol. 45, no. 2, pp. 185–192, 2017.

[26] J. M. Anderson, “Mechanisms of inflammation and
infection with implanted devices,” Cardiovascular
Pathol., vol. 2, no. 3, pp. 33–41, 1993.

[27] T. Someya, Z. Bao, and G. G. Malliaras, “The rise
of plastic bioelectronics,” Nature, vol. 540,
no. 7633, pp. 379–385, 2016.

[28] C. K. Nguyen, N. Q. Tran, T. P. Nguyen, and
D. H. Nguyen, “Biocompatible nanomaterials
based on dendrimers, hydrogels and hydrogel
nanocomposites for use in biomedicine,” Adv.
Natural Sci., Nanosci. Nanotechnol., vol. 8, no. 1,
2017, Art. no. 015001.

[29] K. Wiles, J. M. Fishman, P. De Coppi, and
M. A. Birchall, “The host immune response to
tissue-engineered organs: Current problems and
future directions,” Tissue Eng. B, Rev., vol. 22,
no. 3, pp. 208–219, 2016.

[30] D. Paul, G. Pandey, and R. K. Jain, “Suicidal
genetically engineered microorganisms for
bioremediation: Need and perspectives,”
Bioessays, vol. 27, no. 5, pp. 563–573, 2005.

[31] T. M. Allen and P. R. Cullis, “Drug delivery
systems: Entering the mainstream,” Science,
vol. 303, no. 5665, pp. 1818–1822, 2004.

[32] D. T. Simon et al., “Organic electronics for precise
delivery of neurotransmitters to modulate
mammalian sensory function,” Nature Mater.,
vol. 8, no. 9, pp. 742–746, 2009.

[33] T. S. Moon, C. Lou, A. Tamsir, B. C. Stanton, and
C. A. Voigt, “Genetic programs constructed from
layered logic gates in single cells,” Nature,
vol. 491, no. 7423, pp. 249–253, 2012.

[34] O. B. Akan, O. Cetinkaya, C. Koca, and M. Ozger,
“Internet of hybrid energy harvesting things,” IEEE
Internet Things J., vol. 5, no. 2, pp. 736–746,
Apr. 2018.

[35] C. Dagdeviren, Z. Li, and Z. L. Wang, “Energy
harvesting from the animal/human body for
self-powered electronics,” Annu. Rev. Biomed.
Eng., vol. 19, pp. 85–108, Jun. 2017.

[36] V. Leonov, “Thermoelectric energy harvesting of
human body heat for wearable sensors,” IEEE
Sensors J., vol. 13, no. 6, pp. 2284–2291,
Jun. 2013.

[37] M. A. Karami and D. J. Inman, “Powering
pacemakers from heartbeat vibrations using linear
and nonlinear energy harvesters,” Appl. Phys.
Lett., vol. 100, no. 4, 2012, Art. no. 042901.

[38] W. Jia, G. Valdés-Ramírez, A. J. Bandodkar,
J. R. Windmiller, and J. Wang, “Epidermal biofuel

cells: Energy harvesting from human
perspiration,” Angew. Chem. Int. Ed., vol. 52,
no. 28, pp. 7233–7236, 2013.

[39] S. Hiyama and Y. Moritani, “Molecular
communication: Harnessing biochemical materials
to engineer biomimetic communication systems,”
Nano Commun. Netw., vol. 1, no. 1, pp. 20–30,
May 2010.

[40] C. J. Donnelly, M. Fainzilber, and J. L. Twiss,
“Subcellular communication through RNA
transport and localized protein synthesis,” Traffic,
vol. 11, no. 12, pp. 1498–1505, 2010.

[41] M. Mittelbrunn and F. Sánchez-Madrid,
“Intercellular communication: Diverse structures
for exchange of genetic information,” Nature Rev.
Mol. Cell Biol., vol. 13, no. 5, pp. 328–335, 2012.

[42] N. A. W. Bell and U. F. Keyser, “Digitally encoded
DNA nanostructures for multiplexed,
single-molecule protein sensing with nanopores,”
Nature Nanotechnol., vol. 11, no. 7, pp. 645–651,
2016.

[43] K. Chen et al., “Ionic current-based mapping of
short sequence motifs in single DNA molecules
using solid-state nanopores,” Nano Lett., vol. 17,
no. 9, pp. 5199–5205, 2017.

[44] E. Slonkina and A. B. Kolomeisky, “Polymer
translocation through a long nanopore,” J. Chem.
Phys., vol. 118, no. 15, pp. 7112–7118, 2003.

[45] N. A. Bell, M. Muthukumar, and U. F. Keyser,
“Translocation frequency of double-stranded DNA
through a solid-state nanopore,” Phys. Rev. E, Stat.
Phys. Plasmas Fluids Relat. Interdiscip. Top.,
vol. 93, no. 2, 2016, Art. no. 022401.

[46] N. A. W. Bell and U. F. Keyser. (2016). “Direct
measurements reveal non-Markovian fluctuations
of DNA threading through a solid-state nanopore.”
[Online]. Available:
https://arxiv.org/abs/1607.04612

[47] T. Z. Butler, J. H. Gundlach, and M. A. Troll,
“Determination of RNA orientation during
translocation through a biological nanopore,”
Biophys. J., vol. 90, no. 1, pp. 190–199, 2006.

[48] C. Dekker, “Solid-state nanopores,” Nature
Nanotechnol., vol. 2, no. 4, pp. 209–215, 2007.

[49] S. Hernández-Ainsa and U. F. Keyser, “DNA
origami nanopores: Developments, challenges and
perspectives,” Nanoscale, vol. 6, no. 23,
pp. 14121–14132, 2014.

[50] Y. Wang, D. Zheng, Q. Tan, M. X. Wang, and
L.-Q. Gu, “Nanopore-based detection of circulating
microRNAs in lung cancer patients,” Nature
Nanotechnol., vol. 6, no. 10, pp. 668–674, 2011.

[51] B. A. Bilgin, E. Dinc, and O. B. Akan, “DNA-based
molecular communications,” IEEE Access, vol. 6,
pp. 73119–73129, 2018.

[52] L. Organick et al., “Random access in large-scale
DNA data storage,” Nature Biotechnol., vol. 36,
no. 3, pp. 242–248, 2018.

[53] L. C. Cobo and I. F. Akyildiz, “Bacteria-based
communication in nanonetworks,” Nano Commun.
Netw., vol. 1, no. 4, pp. 244–256,
2010.

[54] S. L. Shipman, J. Nivala, J. D. Macklis, and
G. M. Church, “CRISPR–Cas encoding of a digital
movie into the genomes of a population of living
bacteria,” Nature, vol. 547, no. 7663,
pp. 345–349, 2017.

[55] D. E. Clapham, “Calcium signaling,” Cell, vol. 131,
no. 6, pp. 1047–1058, Dec. 2007.

[56] J. Isaksson, P. Kjäll, D. Nilsson, N. Robinson,
M. Berggren, and A. Richter-Dahlfors, “Electronic
control of Ca2+ signalling in neuronal cells using
an organic electronic ion pump,” Nature Mater.,
vol. 6, no. 9, pp. 673–679, 2007.

[57] T. Nakano, T. Suda, M. Moore, R. Egashira,
A. Enomoto, and K. Arima, “Molecular
communication for nanomachines using
intercellular calcium signaling,” in Proc. 5th IEEE
Conf. Nanotechnol., Jul. 2005, pp. 478–481.

[58] I. Uguz et al., “A microfluidic ion pump for in vivo
drug delivery,” Adv. Mater., vol. 29, no. 27, 2017,
Art. no. 1701217.

[59] A. Jonsson, T. A. Sjöström, K. Tybrandt,
M. Berggren, and D. T. Simon, “Chemical delivery
array with millisecond neurotransmitter release,”

Sci. Adv., vol. 2, no. 11, 2016, Art. no. 1601340.
[60] M. C. S. Lee, E. A. Miller, J. Goldberg, L. Orci, and

R. Schekman, “Bi-directional protein transport
between the ER and golgi,” Annu. Rev. Cell Dev.
Biol., vol. 20, pp. 87–123, Nov. 2004.

[61] K. Denzer, M. J. Kleijmeer, H. F. Heijnen,
W. Stoorvogel, and H. J. Geuze, “Exosome: From
internal vesicle of the multivesicular body to
intercellular signaling device,” J. Cell Sci.,
vol. 113, no. 19, pp. 3365–3374, 2000.

[62] S. R. Schwarze, A. Ho, A. Vocero-Akbani, and
S. F. Dowdy, “In vivo protein transduction: Delivery
of a biologically active protein into the mouse,”
Science, vol. 285, no. 5433, pp. 1569–1572, 1999.

[63] F. Walsh, S. Balasubramaniam, D. Botvich, and
W. Donnelly, “Synthetic protocols for nano sensor
transmitting platforms using enzyme and DNA
based computing,” Nano Commun. Netw., vol. 1,
no. 1, pp. 50–62, 2010.

[64] M. K. Chourasia and S. K. Jain, “Pharmaceutical
approaches to colon targeted drug delivery
systems,” J. Pharm Pharm Sci., vol. 6, no. 1,
pp. 33–66, 2003.

[65] K. Cho et al., “Therapeutic nanoparticles for drug
delivery in cancer,” Clin. Cancer Res., vol. 14,
no. 5, pp. 1310–1316, 2008.

[66] M. C. S. Kuran, H. B. Yilmaz, T. Tugcu, and
I. F. Akyildiz, “Interference effects on modulation
techniques in diffusion based nanonetworks,”
Nano Commun. Netw., vol. 3, no. 1, pp. 65–73,
Mar. 2012.

[67] N.-R. Kim and C.-B. Chae, “Novel modulation
techniques using isomers as messenger molecules
for nano communication networks via diffusion,”
IEEE J. Sel. Areas Commun., vol. 31, no. 12,
pp. 847–856, Dec. 2013.

[68] N. Farsad, W. Guo, and A. W. Eckford, “Tabletop
molecular communication: Text messages through
chemical signals,” PLoS ONE, vol. 8, no. 12, 2013,
Art. no. e82935.

[69] M. Prakash and N. Gershenfeld, “Microfluidic
bubble logic,” Science, vol. 315, no. 5813,
pp. 832–835, Feb. 2007.

[70] M. J. Fuerstman, P. Garstecki, and
G. M. Whitesides, “Coding/decoding and
reversibility of droplet trains in microfluidic
networks,” Science, vol. 315, no. 5813,
pp. 828–832, 2007.

[71] S. L. Peterson, A. McDonald, P. L. Gourley, and
D. Y. Sasaki, “Poly(dimethylsiloxane) thin films as
biocompatible coatings for microfluidic devices:
Cell culture and flow studies with glial cells,”
J. Biomed. Mater. Res. A, Official J. Soc.
Biomaterials, Jpn. Soc. Biomaterials, Austral. Soc.
Biomaterials Korean Soc. Biomaterials, vol. 72A,
no. 1, pp. 10–18, 2005.

[72] N. Farsad, A. W. Eckford, S. Hiyama, and
Y. Moritani, “On-chip molecular communication:
Analysis and design,” IEEE Trans. Nanobiosci.,
vol. 11, no. 3, pp. 304–314, Sep. 2012.

[73] E. De Leo, L. Donvito, L. Galluccio, A. Lombardo,
G. Morabito, and L. M. Zanoli, “Communications
and switching in microfluidic systems: Pure
hydrodynamic control for networking
labs-on-a-chip,” IEEE Trans. Commun., vol. 61,
no. 11, pp. 4663–4677, Nov. 2013.

[74] A. Scott, K. Weir, C. Easton, W. Huynh,
W. J. Moody, and A. Folch, “A microfluidic
microelectrode array for simultaneous
electrophysiology, chemical stimulation, and
imaging of brain slices,” Lab Chip, vol. 13, no. 4,
pp. 527–535, 2013.

[75] R. Iezzi, P. Finlayson, Y. Xu, and R. Katragadda,
“Microfluidic neurotransmiter-based neural
interfaces for retinal prosthesis,” in Proc. IEEE
Annu. Int. Conf. Eng. Med. Biol. Soc. (EMBC),
Sep. 2009, pp. 4563–4565.

[76] A. Jonsson et al., “Bioelectronic neural pixel:
Chemical stimulation and electrical sensing at the
same site,” Proc. Nat. Acad. Sci. USA, vol. 113,
no. 34, pp. 9440–9445, 2016.

[77] P. D. Jones and M. Stelzle, “Can nanofluidic
chemical release enable fast, high resolution
neurotransmitter-based neurostimulation?”
Frontiers Neurosci., vol. 10, p. 138, Mar. 2016.

1336 PROCEEDINGS OF THE IEEE | Vol. 107, No. 7, July 2019



Kuscu et al.: Transmitter and Receiver Architectures for MCs

[78] M. I. Walker et al., “Extrinsic cation selectivity of
2D membranes,” ACS Nano, vol. 11, no. 2,
pp. 1340–1346, 2017.

[79] A. Fabbro et al., “Graphene-based interfaces do
not alter target nerve cells,” ACS Nano, vol. 10,
no. 1, pp. 615–623, 2016.

[80] S. Murdan, “Electro-responsive drug delivery from
hydrogels,” J. Controlled Release, vol. 92, nos. 1–2,
pp. 1–17, 2003.

[81] Y. Brudno et al., “Refilling drug delivery depots
through the blood,” Proc. Nat. Acad. Sci. USA,
vol. 111, no. 35, pp. 12722–12727, 2014.

[82] Y. Brudno et al., “Replenishable drug depot to
combat post-resection cancer recurrence,”
Biomaterials, vol. 178, pp. 373–382, Sep. 2018.

[83] W. Whyte et al., “Sustained release of targeted
cardiac therapy with a replenishable implanted
epicardial reservoir,” Nature Biomed. Eng., vol. 2,
no. 6, pp. 416–428, 2018.

[84] D. Hanahan, “Studies on transformation of
Escherichia coli with plasmids,” J. Mol. Biol.,
vol. 166, no. 4, pp. 557–580, 1983.

[85] G. J. Hannon, “RNA interference,” Nature,
vol. 418, no. 6894, pp. 244–251, 2002.

[86] J. L. Mellies and E. R. Lawrence-Pine,
“Interkingdom signaling between pathogenic
bacteria and Caenorhabditis elegans,” Trends
Microbiol., vol. 18, no. 10, pp. 448–454, 2010.

[87] S. Xiang, J. Fruehauf, and C. J. Li, “Short hairpin
RNA–expressing bacteria elicit RNA interference
in mammals,” Nature Biotechnol., vol. 24, no. 6,
pp. 697–702, 2006.

[88] D. T. Riglar and P. A. Silver, “Engineering bacteria
for diagnostic and therapeutic applications,”
Nature Rev. Microbiol., vol. 16, no. 4, pp. 214–225,
2018.

[89] R. McKay et al., “A platform of genetically
engineered bacteria as vehicles for localized
delivery of therapeutics: Toward applications for
crohn’s disease,” Bioeng. Transl. Med., vol. 3,
no. 3, pp. 209–221, 2018.

[90] M. Gregori, I. Llatser, A. Cabellos-Aparicio, and
E. Alarcón, “Physical channel characterization for
medium-range nanonetworks using flagellated
bacteria,” Comput. Netw., vol. 55, no. 3,
pp. 779–791, 2011.

[91] A. M. Sugrañes and I. F. Akyildiz, “Capacity and
delay of bacteria-based communication in
nanonetworks,” Ph.D. dissertation, Dept.
d’Arquitectura de Computadors, Escola Tècnica
Superior d’Enginyeria de Telecomunicació de
Barcelona, Polytechnic Univ. Catalonia, Barcelona,
Spain, 2012.

[92] P. Liò and S. Balasubramaniam, “Opportunistic
routing through conjugation in bacteria
communication nanonetwork,” Nano Commun.
Netw., vol. 3, no. 1, pp. 36–45, 2012.

[93] S. Balasubramaniam and P. Liò, “Multi-hop
conjugation based bacteria nanonetworks,” IEEE
Trans. Nanobiosci., vol. 12, no. 1, pp. 47–59,
Mar. 2013.

[94] V. Petrov et al., “Forward and reverse coding for
chromosome transfer in bacterial nanonetworks,”
Nano Commun. Netw., vol. 5, nos. 1–2, pp. 15–24,
2014.

[95] V. Petrov, B. Deng, D. Moltchanov,
S. Balasubramaniam, and Y. Koucheryavy,
“Performance comparison of message encoding
techniques for bacterial nanonetworks,” in Proc.
IEEE Wireless Commun. Netw. Conf. (WCNC),
Apr. 2016, pp. 1–7.

[96] P. Palese, H. Zheng, O. G. Engelhardt, S. Pleschka,
and A. García-Sastre, “Negative-strand RNA
viruses: Genetic engineering and applications,”
Proc. Nat. Acad. Sci. USA, vol. 93, no. 21,
pp. 11354–11358, 1996.

[97] C. E. Dunbar, K. A. High, J. K. Joung, D. B. Kohn,
K. Ozawa, and M. Sadelain, “Gene therapy comes
of age,” Science, vol. 359, no. 6372, p. eaan4672,
2018.

[98] J. Yu, J. Moffitt, C. L. Hetherington,
C. Bustamante, and G. Oster, “Mechanochemistry
of a viral DNA packaging motor,” J. Mol. Biol.,
vol. 400, no. 2, pp. 186–203, 2010.

[99] F. Walsh and S. Balasubramaniam, “Reliability of

multi-path virus nanonetworks,” in Proc. IEEE Int.
Conf. Commun. Workshops (ICC), Jun. 2013,
pp. 824–828.

[100] F. Walsh and S. Balasubramaniam, “Reliability and
delay analysis of multihop virus-based
nanonetworks,” IEEE Trans. Nanotechnol., vol. 12,
no. 5, pp. 674–684, Sep. 2013.

[101] S. Mathivanan, H. Ji, and R. J. Simpson,
“Exosomes: Extracellular organelles important in
intercellular communication,” J. Proteomics,
vol. 73, no. 10, pp. 1907–1920, 2010.

[102] S. Abadal and I. F. Akyildiz, “Automata modeling
of quorum sensing for nanocommunication
networks,” Nano Commun. Netw., vol. 2, no. 1,
pp. 74–83, 2011.

[103] S. Abadal, I. Llatser, E. Alarcón, and
A. Cabellos-Aparicio, “Quorum sensing-enabled
amplification for molecular nanonetworks,” in
Proc. IEEE Int. Conf. Commun. (ICC), Jun. 2012,
pp. 6162–6166.

[104] D. P. Martins, K. Leetanasaksakul, M. T. Barros,
A. Thamchaipenet, W. Donnelly, and
S. Balasubramaniam, “Molecular communications
pulse-based jamming model for bacterial biofilm
suppression,” IEEE Trans. Nanobiosci., vol. 17,
no. 4, pp. 533–542, Oct. 2018.

[105] T. Nakano, M. J. Moore, F. Wei, A. V. Vasilakos,
and J. Shuai, “Molecular communication and
networking: Opportunities and challenges,” IEEE
Trans. Nanobiosci., vol. 11, no. 2, pp. 135–148,
Jun. 2012.

[106] M. Brini, T. Calì, D. Ottolini, and E. Carafoli,
“Intracellular calcium homeostasis and signaling,”
in Metallomics and the Cell. Dordrecht,
The Netherlands: Springer, 2013,
pp. 119–168.

[107] M. L. Cotrina et al., “Connexins regulate calcium
signaling by controlling ATP release,” Proc. Nat.
Acad. Sci. USA, vol. 95, no. 26, pp. 15735–15740,
1998.

[108] T. Nakano, T. Suda, T. Koujin, T. Haraguchi, and
Y. Hiraoka, “Molecular communication through
gap junction channels: System design,
experiments and modeling,” in Proc. IEEE 2nd Inf.
Comput. Syst. Bio-Inspired Models Netw.
(Bionetics), Dec. 2007, pp. 139–146.

[109] T. Nakano and J.-Q. Liu, “Design and analysis of
molecular relay channels: An information
theoretic approach,” IEEE Trans. Nanobiosci.,
vol. 9, no. 3, pp. 213–221, Sep. 2010.

[110] M. T. Barros, S. Balasubramaniam, B. Jennings,
and Y. Koucheryavy, “Transmission protocols for
calcium-signaling-based molecular
communications in deformable cellular tissue,”
IEEE Trans. Nanotechnol., vol. 13, no. 4,
pp. 779–788, Jul. 2014.

[111] M. T. Barros, S. Balasubramaniam, and
B. Jennings, “Using information metrics and
molecular communication to detect cellular tissue
deformation,” IEEE Trans. Nanobiosci., vol. 13,
no. 3, pp. 278–288, Sep. 2014.

[112] M. T. Barros, S. Balasubramaniam, and
B. Jennings, “Comparative end-to-end analysis of
Ca2+-signaling-based molecular communication
in biological tissues,” IEEE Trans. Commun.,
vol. 63, no. 12, pp. 5128–5142, Dec. 2015.

[113] M. T. Barros, “Ca2+ -signaling-based molecular
communication systems: Design and future
research directions,” Nano Commun. Netw.,
vol. 11, pp. 103–113, Mar. 2017.

[114] M. Moore et al., “A design of a molecular
communication system for nanomachines using
molecular motors,” in Proc. 4th Annu. IEEE Int.
Conf. Pervasive Comput. Commun. Workshops
(PerCom Workshops), Mar. 2006, pp. 554–559.

[115] A. Enomoto et al., “A molecular communication
system using a network of cytoskeletal filaments,”
in Proc. NSTI Nanotechnol. Conf., 2006.

[116] A. W. Eckford, N. Farsad, S. Hiyama, and
Y. Moritani, “Microchannel molecular
communication with nanoscale carriers: Brownian
motion versus active transport,” in Proc. 10th IEEE
Conf. Nanotechnol. (IEEE-NANO), Aug. 2010,
pp. 854–858.

[117] A. Enomoto, M. J. Moore, T. Suda, and K. Oiwa,

“Design of self-organizing microtubule networks
for molecular communication,” Nano Commun.
Netw., vol. 2, no. 1, pp. 16–24, Mar. 2011.

[118] N. Farsad, A. W. Eckford, and S. Hiyama, “Design
and optimizing of on-chip kinesin substrates for
molecular communication,” IEEE Trans.
Nanotechnol., vol. 14, no. 4, pp. 699–708,
Jul. 2015.

[119] S. Sankaran, J. Becker, C. Wittmann, and
A. del Campo, “Optoregulated drug release from
an engineered living hydrogel,” Tech. Rep., 2018.

[120] S. R. Lindemann et al., “Engineering microbial
consortia for controllable outputs,” ISME J.,
vol. 10, no. 9, pp. 2077–2084, 2016.

[121] M. Kuscu and O. B. Akan, “On the physical design
of molecular communication receiver based on
nanoscale biosensors,” IEEE Sensors J., vol. 16,
no. 8, pp. 2228–2243, Sep. 2016.

[122] D. A. Armbruster and T. Pry, “Limit of blank, limit
of detection and limit of quantitation,” Clin.
Biochemist Rev., vol. 29, pp. S49–S52, Aug. 2008.

[123] P. R. Nair and M. A. Alam, “Design considerations
of silicon nanowire biosensors,” IEEE Trans.
Electron Devices, vol. 54, no. 12, pp. 3400–3408,
Dec. 2007.

[124] K. Kalantar-Zadeh, Sensors: An Introductory
Course. New York, NY, USA: Springer, 2013.

[125] R. Daniel, J. R. Rubens, R. Sarpeshkar, and
T. K. Lu, “Synthetic analog computation in living
cells,” Nature, vol. 497, no. 7451, pp. 619–623,
2013.

[126] O. Purcell and T. K. Lu, “Synthetic analog and
digital circuits for cellular computation and
memory,” Current Opinion Biotechnol., vol. 29,
pp. 146–155, Oct. 2014.

[127] M. Pierobon, “A systems-theoretic model of a
biological circuit for molecular communication in
nanonetworks,” Nano Commun. Netw., vol. 5,
nos. 1–2, pp. 25–34, Mar./Jun. 2014.

[128] C. J. Myers, Engineering Genetic Circuits. London,
U.K.: Chapman & Hall, 2016.

[129] D. Baker et al., “Engineering life: Building a FAB
for biology,” Sci. Amer., vol. 294, no. 6, pp. 44–51,
2006.

[130] R. Sarpeshkar, “Analog synthetic biology,” Philos.
Trans. Roy. Soc. A, Math. Phys. Eng. Sci., vol. 372,
no. 2012, 2014, Art. no. 20130110.

[131] A. Marcone, M. Pierobon, and M. Magarini,
“Parity-check coding based on genetic circuits for
engineered molecular communication between
biological cells,” IEEE Trans. Commun., vol. 66,
no. 12, pp. 6221–6236, Dec. 2018.

[132] N. Arkus, “A mathematical model of cellular
apoptosis and senescence through the dynamics of
telomere loss,” J. Theor. Biol., vol. 235, no. 1,
pp. 13–32, 2005.

[133] P. K. Maini, “Using mathematical models to help
understand biological pattern formation,” Comp.
Rendus Biol., vol. 327, no. 3, pp. 225–234, 2004.

[134] E. Klipp and W. Liebermeister, “Mathematical
modeling of intracellular signaling pathways,”
BMC Neurosci., vol. 7, no. 1, p. S10, Oct. 2006.

[135] C. Harper, M. Pierobon, and M. Magarini,
“Estimating information exchange performance of
engineered cell-to-cell molecular communications:
A computational approach,” in Proc. IEEE
INFOCOM Conf. Comput. Commun., Apr. 2018,
pp. 729–737.

[136] T. Ramalho, A. Meyer, A. Mückl, K. Kapsner,
U. Gerland, and F. C. Simmel, “Single cell analysis
of a bacterial sender-receiver system,” PLoS ONE,
vol. 11, no. 1, 2016, Art. no. e0145829.

[137] M. Kuscu and O. B. Akan, “Modeling and analysis
of SiNW FET-based molecular communication
receiver,” IEEE Trans. Commun., vol. 64, no. 9,
pp. 3708–3721, Sep. 2016.

[138] N. Farsad, D. Pan, and A. Goldsmith, “A novel
experimental platform for in-vessel multi-chemical
molecular communications,” in Proc. GLOBECOM
IEEE Global Commun. Conf., Dec. 2017, pp. 1–6.

[139] H. Unterweger et al. (2018). “Experimental
molecular communication testbed based on
magnetic nanoparticles in duct flow.” [Online].
Available: https://arxiv.org/abs/1803.06990

[140] S. M. Borisov and O. S. Wolfbeis, “Optical

Vol. 107, No. 7, July 2019 | PROCEEDINGS OF THE IEEE 1337



Kuscu et al.: Transmitter and Receiver Architectures for MCs

biosensors,” Chem. Rev., vol. 108, no. 2,
pp. 423–461, Feb. 2008. [Online]. Available:
http://www.ncbi.nlm.nih.gov/pubmed/18229952

[141] J. L. Arlett, E. B. Myers, and M. L. Roukes,
“Comparative advantages of mechanical
biosensors,” Nature Nanotechnol., vol. 6, no. 4,
pp. 203–215, 2011.

[142] J. Wang, “Electrochemical glucose biosensors,”
Chem. Rev., vol. 108, no. 2, pp. 814–825, 2008.

[143] K. R. Rogers, “Principles of affinity-based
biosensors,” Mol. Biotechnol., vol. 14, no. 2,
pp. 109–129, 2000.

[144] Y. Huang, X. Dong, Y. Shi, C. M. Li, L.-J. Li, and
P. Chen, “Nanoelectronic biosensors based on CVD
grown graphene,” Nanoscale, vol. 2, no. 8,
pp. 1485–1488, Aug. 2010.

[145] A. Poghossian and M. J. Schöning, “Label-free
sensing of biomolecules with field-effect devices
for clinical applications,” Electroanalysis, vol. 26,
no. 6, pp. 1197–1213, 2014.

[146] M. F. Bear, B. W. Connors, and M. A. Paradiso,
Neuroscience: Exploring the Brain, vol. 2.
Baltimore, MD, USA: Williams & Wilkins, 2007.

[147] J. Shan et al., “High sensitivity glucose detection
at extremely low concentrations using a
MoS2-based field-effect transistor,” RSC Adv.,
vol. 8, no. 15, pp. 7942–7948, 2018.

[148] S. Xu et al., “Real-time reliable determination of
binding kinetics of DNA hybridization using a
multi-channel graphene biosensor,” Nature
Commun., vol. 8, Mar. 2017, Art. no. 14902.

[149] B. Kim et al., “Highly selective and sensitive
detection of neurotransmitters using
receptor-modified single-walled carbon nanotube
sensors,” Nanotechnology, vol. 24, no. 28, 2013,
Art. no. 285501.

[150] J. P. Kim, B. Y. Lee, J. Lee, S. Hong, and S. J. Sim,
“Enhancement of sensitivity and specificity by
surface modification of carbon nanotubes in
diagnosis of prostate cancer based on carbon
nanotube field effect transistors,” Biosensors
Bioelectron., vol. 24, no. 11, pp. 3372–3378, 2009.

[151] Y.-S. Lo et al., “Oriented immobilization of
antibody fragments on ni-decorated single-walled
carbon nanotube devices,” ACS Nano, vol. 3,
no. 11, pp. 3649–3655, 2009.

[152] F. Patolsky, G. Zheng, and C. M. Lieber,
“Nanowire-based biosensors,” Anal. Chem.,
vol. 78, no. 13, pp. 4260–4269, 2006.

[153] G.-J. Zhang et al., “Highly sensitive and reversible
silicon nanowire biosensor to study nuclear
hormone receptor protein and response element
DNA interactions,” Biosensors Bioelectron., vol. 26,
no. 2, pp. 365–370, 2010.

[154] J. Liu, J. Goud, P. M. Raj, M. Iyer, Z. Wang, and
R. R. Tummala, “Label-free protein detection by
zno nanowire based bio-sensors,” in Proc. IEEE
57th Electron. Compon. Technol. Conf. (ECTC),
May/Jun. 2007, pp. 1971–1976.

[155] W. Fu et al., “Biosensing near the neutrality point
of graphene,” Sci. Adv., vol. 3, no. 10, 2017,
Art. no. e1701247.

[156] Y. Ohno, K. Maehashi, and K. Matsumoto,
“Label-free biosensors based on aptamer-modified
graphene field-effect transistors,” J. Amer. Chem.
Soc., vol. 132, no. 51, pp. 18012–18013, 2010.

[157] H. S. Song et al., “Bioelectronic tongue using
heterodimeric human taste receptor for the
discrimination of sweeteners with human-like
performance,” ACS Nano, vol. 8, no. 10,
pp. 9781–9789, 2014.

[158] C.-M. Lim, J. Y. Kwon, and W.-J. Cho, “Field-effect
transistor biosensor platform fused with
Drosophila odorant-binding proteins for instant
ethanol detection,” ACS Appl. Mater. Interfaces,
vol. 9, no. 16, pp. 14051–14057, 2017.

[159] S. Eissa and M. Zourob, “In vitro selection of DNA
aptamers targeting β-lactoglobulin and their
integration in graphene-based biosensor for the
detection of milk allergen,” Biosensors Bioelectron.,
vol. 91, pp. 169–174, May 2017.

[160] E. Luzi, M. Minunni, S. Tombelli, and M. Mascini,
“New trends in affinity sensing: Aptamers for
ligand binding,” TrAC Trends Anal. Chem., vol. 22,
no. 11, pp. 810–818, 2003.

[161] W. Yang, K. R. Ratinac, S. P. Ringer, P. Thordarson,
J. J. Gooding, and F. Braet, “Carbon nanomaterials
in biosensors: Should you use nanotubes or
graphene?” Angew. Chem. Int. Ed., vol. 49, no. 12,
pp. 2114–2138, 2010.

[162] L. Torsi, M. Magliulo, K. Manoli, and G. Palazzo,
“Organic field-effect transistor sensors: A tutorial
review,” Chem. Soc. Rev., vol. 42, no. 22,
pp. 8612–8628, 2013.

[163] M. Curreli et al., “Real-time, label-free detection
of biological entities using nanowire-based FETs,”
IEEE Trans. Nanotechnol., vol. 7, no. 6,
pp. 651–667, Nov. 2008.

[164] S. U. Senveli and O. Tigli, “Biosensors in the small
scale: Methods and technology trends,” IET
Nanobiotechnol., vol. 7, no. 1, pp. 7–21,
Mar. 2013.

[165] D. P. Tran et al., “Toward intraoperative detection
of disseminated tumor cells in lymph nodes with
silicon nanowire field effect transistors,” ACS
Nano, vol. 10, no. 2, pp. 2357–2364, Feb. 2016.

[166] X. Duan, Y. Li, N. K. Rajan, D. A. Routenberg,
Y. Modis, and M. A. Reed, “Quantification of the
affinities and kinetics of protein interactions using
silicon nanowire biosensors,” Nature Nanotechnol.,
vol. 7, no. 6, pp. 401–407, 2012.

[167] K.-I. Chena, B.-R. Li, and Y.-T. Chen, “Silicon
nanowire field-effect transistor-based biosensors
for biomedical diagnosis and cellular recording
investigation,” Nano Today, vol. 6, no. 2,
pp. 131–154, 2011.

[168] B. Zhan, C. Li, J. Yang, G. Jenkins, W. Huang, and
X. Dong, “Graphene field-effect transistor and its
application for electronic sensing,” Small, vol. 10,
no. 20, pp. 4042–4065, 2014.

[169] G.-J. Zhang and Y. Ning, “Silicon nanowire
biosensor and its applications in disease
diagnostics: A review,” Anal. Chim. Acta, vol. 749,
pp. 1–15, Oct. 2012.

[170] D. P. Tran, T. T. Pham, B. Wolfrum,
A. Offenhäusser, and B. Thierry,
“CMOS-compatible silicon nanowire field-effect
transistor biosensor: Technology development
toward commercialization,” Materials, vol. 11,
no. 5, p. 785, 2018.

[171] K. Maehashi and K. Matsumoto, “Label-free
electrical detection using carbon nanotube-based
biosensors,” Sensors, vol. 9, no. 7, pp. 5368–5378,
2009.

[172] S. K. Smart, A. I. Cassady, G. Q. Lu, and
D. J. Martin, “The biocompatibility of carbon
nanotubes,” Carbon, vol. 44, no. 6,
pp. 1034–1047, 2006.

[173] S. Kumar, R. Rani, N. Dilbaghi, K. Tankeshwar,
and K.-H. Kim, “Carbon nanotubes: A novel
material for multifaceted applications in human
healthcare,” Chem. Soc. Rev., vol. 46, no. 1,
pp. 158–196, 2017.

[174] M. Sireesha, V. J. Babu, A. S. K. Kiran, and
S. Ramakrishna, “A review on carbon nanotubes in
biosensor devices and their applications in
medicine,” Nanocomposites, vol. 4, no. 2,
pp. 36–57, 2018.

[175] S. J. Park, O. S. Kwon, S. H. Lee, H. S. Song,
T. H. Park, and J. Jang, “Ultrasensitive flexible
graphene based field-effect transistor (FET)-type
bioelectronic nose,” Nano Lett., vol. 12, no. 10,
pp. 5082–5090, 2012.

[176] J. Ping, R. Vishnubhotla, A. Vrudhula, and
A. T. C. Johnson, “Scalable production of
high-sensitivity, label-free DNA biosensors based
on back-gated graphene field effect transistors,”
ACS Nano, vol. 10, no. 9, pp. 8700–8704, 2016.

[177] N. Gao et al., “Specific detection of biomolecules
in physiological solutions using graphene
transistor biosensors,” Proc. Nat. Acad. Sci. USA,
vol. 113, no. 51, pp. 14633–14638, 2016.

[178] A. K. Manoharan, S. Chinnathambi, R. Jayavel,
and N. Hanagata, “Simplified detection of the
hybridized DNA using a graphene field effect
transistor,” Sci. Technol. Adv. Mater., vol. 18, no. 1,
pp. 43–50, 2017.

[179] G. Wu et al., “Graphene field-effect transistors for
the sensitive and selective detection of escherichia
coli using pyrene-tagged DNA aptamer,” Adv.

Healthcare Mater., vol. 6, no. 19, 2017,
Art. no. 1700736.

[180] S. Viswanathan et al., “Graphene–protein field
effect biosensors: Glucose sensing,” Mater. Today,
vol. 18, no. 9, pp. 513–522, 2015.

[181] A. Hassibi, S. Zahedi, R. Navid, R. W. Dutton, and
T. H. Lee, “Biological shot-noise and
quantum-limited signal-to-noise ratio in
affinity-based biosensors,” J. Appl. Phys., vol. 97,
no. 8, 2005, Art. no. 084701.

[182] A. M. Berezhkovskii and A. Szabo, “Effect of
ligand diffusion on occupancy fluctuations of
cell-surface receptors,” J. Chem. Phys., vol. 139,
no. 12, 2013, Art. no. 121910.

[183] M. Pierobon and I. F. Akyildiz, “Noise analysis in
ligand-binding reception for molecular
communication in nanonetworks,” IEEE Trans.
Signal Process., vol. 59, no. 9, pp. 4168–4182,
Sep. 2011.

[184] M. Pierobon and I. F. Akyildiz,
“A statistical–physical model of interference in
diffusion-based molecular nanonetworks,” IEEE
Trans. Commun., vol. 62, no. 6, pp. 2085–2095,
Jun. 2014.

[185] M. Pierobon and I. F. Akyildiz, “Intersymbol and
co-channel interference in diffusion-based
molecular communication,” in Proc. IEEE Int. Conf.
Commun. (ICC), Jun. 2012, pp. 6126–6131.

[186] M. J. Deen, M. W. Shinwari, J. C. Ranuárez, and
D. Landheer, “Noise considerations in field-effect
biosensors,” J. Appl. Phys., vol. 100, no. 7, 2006,
Art. no. 074703.

[187] N. K. Rajan, X. Duan, and M. A. Reed,
“Performance limitations for
nanowire/nanoribbon biosensors,” Wiley
Interdiscipl. Rev., Nanomed. Nanobiotechnol.,
vol. 5, no. 6, pp. 629–645, 2013.

[188] Y. Kutovyi et al., “Origin of noise in liquid-gated Si
nanowire troponin biosensors,” Nanotechnology,
vol. 29, no. 17, 2018, Art. no. 175202.

[189] M. Khalid, O. Amin, S. Ahmed, and M.-S. Alouini,
“System modeling of virus transmission and
detection in molecular communication channels,”
in Proc. IEEE Int. Conf. Commun. (ICC), May 2018,
pp. 1–6. [Online]. Available:
https://ieeexplore.ieee.org/document/8422665/

[190] H. Hibino, H. Kageshima, M. Kotsugi, F. Maeda,
F.-Z. Guo, and Y. Watanabe, “Dependence of
electronic properties of epitaxial few-layer
graphene on the number of layers investigated by
photoelectron emission microscopy,” Phys. Rev. B,
Condens. Matter, vol. 79, no. 12, 2009,
Art. no. 125437.

[191] H. M. Wang, Y. H. Wu, Z. H. Ni, and Z. X. Shen,
“Electronic transport and layer engineering in
multilayer graphene structures,” Appl. Phys. Lett.,
vol. 92, no. 5, 2008, Art. no. 053504.

[192] J.-H. Chen, C. Jang, S. Xiao, M. Ishigami, and
M. S. Fuhrer, “Intrinsic and extrinsic performance
limits of graphene devices on SiO2,” Nature
Nanotechnol., vol. 3, no. 4, pp. 206–209, 2008.

[193] K. I. Bolotin et al., “Ultrahigh electron mobility in
suspended graphene,” Solid State Commun.,
vol. 146, pp. 351–355, Jun. 2008.

[194] A. H. C. Neto, F. Guinea, N. M. R. Peres,
K. S. Novoselov, and A. K. Geim, “The electronic
properties of graphene,” Rev. Mod. Phys., vol. 81,
no. 1, p. 109, Jan. 2009.

[195] T. Enoki, Y. Kobayashi, and K.-I. Fukui, “Electronic
structures of graphene edges and nanographene,”
Int. Rev. Phys. Chem., vol. 26, no. 4, pp. 609–645,
2007.

[196] V. B. Shenoy, C. D. Reddy, A. Ramasubramaniam,
and Y. W. Zhang, “Edge-stress-induced warping of
graphene sheets and nanoribbons,” Phys. Rev.
Lett., vol. 101, no. 24, 2008, Art. no. 245501.

[197] V. Geringer et al., “Intrinsic and extrinsic
corrugation of monolayer graphene deposited on
SiO2 ,” Phys. Rev. Lett., vol. 102, no. 7, 2009,
Art. no. 076102.

[198] L. Brey and J. J. Palacios, “Exchange-induced
charge inhomogeneities in rippled neutral
graphene,” Phys. Rev. B, Condens. Matter, vol. 77,
no. 4, 2008, Art. no. 041403.

[199] Z. Ni, Y. Wang, T. Yu, Y. You, and Z. Shen,

1338 PROCEEDINGS OF THE IEEE | Vol. 107, No. 7, July 2019



Kuscu et al.: Transmitter and Receiver Architectures for MCs

“Reduction of Fermi velocity in folded graphene
observed by resonance Raman spectroscopy,”
Phys. Rev. B, Condens. Matter, vol. 77, no. 23,
2008, Art. no. 235403.

[200] N.-R. Kim, N. Farsad, C.-B. Chae, and
A. W. Eckford, “A universal channel model for
molecular communication systems with
metal-oxide detectors,” in Proc. IEEE Int. Conf.
Commun. (ICC), Jun. 2015, pp. 1054–1059.

[201] B.-H. Koo, C. Lee, H. B. Yilmaz, N. Farsad,
A. Eckford, and C.-B. Chae, “Molecular MIMO:
From theory to prototype,” IEEE J. Sel. Areas
Commun., vol. 34, no. 3, pp. 600–614, Mar. 2016.

[202] W. Wicke, A. Ahmadzadeh, V. Jamali,
H. Unterweger, C. Alexiou, and R. Schober.
(2018). “Magnetic nanoparticle based molecular
communication in microfluidic environments.”
[Online]. Available:
https://arxiv.org/abs/1808.05147

[203] M. U. Mahfuz, D. Makrakis, and H. T. Mouftah,
“On the characterization of binary
concentration-encoded molecular communication
in nanonetworks,” Nano Commun. Netw., vol. 1,
no. 4, pp. 289–300, Dec. 2010.

[204] N. Garralda, I. Llatser, A. Cabellos-Aparicio,
E. Alarcón, and M. Pierobon, “Diffusion-based
physical channel identification in molecular
nanonetworks,” Nano Commun. Netw., vol. 2,
no. 4, pp. 196–204, 2011.

[205] M. H. Kabir, S. M. R. Islam, and K. S. Kwak,
“D-MoSK modulation in molecular
communications,” IEEE Trans. Nanobiosci., vol. 14,
no. 6, pp. 680–683, Sep. 2015.

[206] Y. Murin, N. Farsad, M. Chowdhury, and
A. Goldsmith, “Communication over
diffusion-based molecular timing channels,” in
Proc. IEEE Global Commun. Conf. (GLOBECOM),
Dec. 2016, pp. 1–6.

[207] K. V. Srinivas, A. W. Eckford, and R. S. Adve,
“Molecular communication in fluid media: The
additive inverse Gaussian noise channel,” IEEE
Trans. Inf. Theory, vol. 58, no. 7, pp. 4678–4692,
Jul. 2012.

[208] Y. Murin, N. Farsad, M. Chowdhury, and
A. Goldsmith. (2018). “Exploiting diversity in
molecular timing channels via order statistics.”
[Online]. Available:
https://arxiv.org/abs/1801.05567

[209] B. Atakan, S. Galmes, and O. B. Akan, “Nanoscale
communication with molecular arrays in
nanonetworks,” IEEE Trans. Nanobiosci., vol. 11,
no. 2, pp. 149–160, Jun. 2012.

[210] M. Movahednasab, M. Soleimanifar, A. Gohari,
M. Nasiri-Kenari, and U. Mitra, “Adaptive
transmission rate with a fixed threshold decoder
for diffusion-based molecular communication,”
IEEE Trans. Commun., vol. 64, no. 1, pp. 236–248,
Jan. 2015.

[211] M. Blawat et al., “Forward error correction for
DNA data storage,” Procedia Comput. Sci., vol. 80,
pp. 1011–1022, 2016.

[212] J. Clarke, H.-C. Wu, L. Jayasinghe, A. Patel,
S. Reid, and H. Bayley, “Continuous base
identification for single-molecule nanopore DNA
sequencing,” Nature Nanotechnol., vol. 4, no. 4,
pp. 265–270, 2009.

[213] I. M. Derrington et al., “Nanopore DNA
sequencing with MspA,” Proc. Nat. Acad. Sci. USA,
vol. 107, no. 37, pp. 16060–16065, 2010.

[214] R. W. Hamming, “Error detecting and error
correcting codes,” Bell Syst. Tech. J., vol. 29, no. 2,
pp. 147–160, Apr. 1950.

[215] C. Berrou, A. Glavieux, and P. Thitimajshima,
“Near Shannon limit error-correcting coding and
decoding: Turbo-codes. 1,” in Proc. IEEE Int. Conf.
Commun. (ICC), vol. 2, Geneva, Switzerland,
May 1993, pp. 1064–1070.

[216] P.-J. Shih, C.-H. Lee, and P.-C. Yeh, “Channel codes
for mitigating intersymbol interference in
diffusion-based molecular communications,” in
Proc. IEEE Global Commun. Conf. (GLOBECOM),
Dec. 2012, pp. 4228–4232.

[217] P.-J. Shih, C.-H. Lee, P.-C. Yeh, and K.-C. Chen,
“Channel codes for reliability enhancement in
molecular communication,” IEEE J. Sel. Areas

Commun., vol. 31, no. 12, pp. 857–867,
Dec. 2013.

[218] P.-Y. Ko, Y.-C. Lee, P.-C. Yeh, C.-H. Lee, and
K.-C. Chen, “A new paradigm for channel coding
in diffusion-based molecular communications:
Molecular coding distance function,” in Proc. IEEE
Global Commun. Conf. (GLOBECOM), Dec. 2012,
pp. 3748–3753.

[219] M. S. Leeson and M. D. Higgins, “Forward error
correction for molecular communications,” Nano
Commun. Netw., vol. 3, no. 3, pp. 161–167, 2012.

[220] C. Bai, M. S. Leeson, and M. D. Higgins,
“Minimum energy channel codes for molecular
communications,” Electron. Lett., vol. 50, no. 23,
pp. 1669–1671, 2014.

[221] Y. Lu, M. D. Higgins, and M. S. Leeson,
“Self-orthogonal convolutional codes (SOCCs) for
diffusion-based molecular communication
systems,” in Proc. IEEE Int. Conf. Commun. (ICC),
Jun. 2015, pp. 1049–1053.

[222] Y. Lu, X. Wang, M. D. Higgins, A. Noel,
N. Neophytou, and M. S. Leeson, “Energy
requirements of error correction codes in
diffusion-based molecular communication
systems,” Nano Commun. Netw., vol. 11,
pp. 24–35, Mar. 2017.

[223] Y. Lu, M. D. Higgins, and M. S. Leeson,
“Comparison of channel coding schemes for
molecular communications systems,” IEEE Trans.
Commun., vol. 63, no. 11, pp. 3991–4001,
Nov. 2015.

[224] M. U. Mahfuz, D. Makrakis, and H. T. Mouftah,
“Performance analysis of convolutional coding
techniques in diffusion-based
concentration-encoded PAM molecular
communication systems,” BioNanoScience, vol. 3,
no. 3, pp. 270–284, 2013.

[225] A. Marcone, M. Pierobon, and M. Magarini,
“A biological circuit design for modulated
parity-check encoding in molecular
communication,” in Proc. IEEE Int. Conf. Commun.
(ICC), May 2017, pp. 1–7.

[226] M. Pierobon and I. F. Akyildiz, “Diffusion-based
noise analysis for molecular communication in
nanonetworks,” IEEE Trans. Signal Process.,
vol. 59, no. 6, pp. 2532–2547, Jun. 2011.

[227] H. B. Yilmaz, A. C. Heren, T. Tugcu, and
C.-B. Chae, “Three-dimensional channel
characteristics for molecular communications with
an absorbing receiver,” IEEE Commun. Lett.,
vol. 18, no. 6, pp. 929–932, Jun. 2014.

[228] A. Noel, K. C. Cheung, and R. Schober, “Improving
receiver performance of diffusive molecular
communication with enzymes,” IEEE Trans.
Nanobiosci., vol. 13, no. 1, pp. 31–43, Mar. 2014.

[229] A. Noel, K. C. Cheung, and R. Schober, “A unifying
model for external noise sources and ISI in
diffusive molecular communication,” IEEE J. Sel.
Areas Commun., vol. 32, no. 12, pp. 2330–2343,
Dec. 2014.

[230] L. S. Meng, P. C. Yeh, K. C. Chen, and I. F. Akyildiz,
“On receiver design for diffusion-based molecular
communication,” IEEE Trans. Signal Process.,
vol. 62, no. 22, pp. 6032–6044, Nov. 2014.

[231] V. Jamali, A. Ahmadzadeh, and R. Schober, “On
the design of matched filters for molecule
counting receivers,” IEEE Commun. Lett., vol. 21,
no. 8, pp. 1711–1714, Aug. 2017.

[232] A. Noel, K. Cheung, and R. Schober, “Optimal
receiver design for diffusive molecular
communication with flow and additive noise,”
IEEE Trans. Nanobiosci., vol. 13, no. 3,
pp. 350–362, Sep. 2014.

[233] A. Shahbazi and A. Jamshidi, “Adaptive weighted
signal detection for nanoscale molecular
communications,” in Proc. Int. Zürich Seminar Inf.
Commun. (IZS). ETH, Zürich, Switzerland, 2018,
pp. 149–152.

[234] P. He, Y. Mao, Q. Liu, and K. Yang, “Improving
reliability performance of diffusion-based
molecular communication with adaptive threshold
variation algorithm,” Int. J. Commun. Syst.,
vol. 29, no. 18, pp. 2669–2680, 2016.

[235] G. H. Alshammri, M. S. Alzaidi, W. K. M. Ahmed,
and V. B. Lawrence, “Low-complexity

memory-assisted adaptive-threshold detection
scheme for on-off-keying diffusion-based
molecular communications,” in Proc. IEEE 38th
Sarnoff Symp., Sep. 2017, pp. 1–6.

[236] M. Damrath and P. A. Hoeher, “Low-complexity
adaptive threshold detection for molecular
communication,” IEEE Trans. Nanobiosci., vol. 15,
no. 3, pp. 200–208, Apr. 2016.

[237] H. ShahMohammadian, G. G. Messier, and
S. Magierowski, “Optimum receiver for molecule
shift keying modulation in diffusion-based
molecular communication channels,” Nano
Commun. Netw., vol. 3, no. 3, pp. 183–195, 2012.

[238] M. U. Mahfuz, D. Makrakis, and H. T. Mouftah,
“A comprehensive analysis of strength-based
optimum signal detection in
concentration-encoded molecular communication
with spike transmission,” IEEE Trans. Nanobiosci.,
vol. 14, no. 1, pp. 67–83, Jan. 2015.

[239] M. U. Mahfuz, D. Makrakis, and H. T. Mouftah,
“A comprehensive study of sampling-based
optimum signal detection in
concentration-encoded molecular
communication,” IEEE Trans. Nanobiosci., vol. 13,
no. 3, pp. 208–222, Sep. 2014.

[240] H. Yan, G. Chang, Z. Ma, and L. Lin,
“Derivative-based signal detection for high data
rate molecular communication system,” IEEE
Commun. Lett., vol. 22, no. 9, pp. 1782–1785,
Sep. 2018.

[241] B. Li, M. Sun, S. Wang, W. Guo, and C. Zhao,
“Low-complexity noncoherent signal detection for
nanoscale molecular communications,” IEEE
Trans. Nanobiosci., vol. 15, no. 1, pp. 3–10,
Jan. 2016.

[242] B. Li, M. Sun, S. Wang, W. Guo, and C. Zhao,
“Local convexity inspired low-complexity
noncoherent signal detector for nanoscale
molecular communications,” IEEE Trans.
Commun., vol. 64, no. 5, pp. 2079–2091,
May 2016.

[243] V. Jamali, N. Farsad, R. Schober, and A. Goldsmith,
“Non-coherent detection for diffusive molecular
communication systems,” IEEE Trans. Commun.,
vol. 66, no. 6, pp. 2515–2531, Jun. 2018.

[244] V. Jamali, A. Ahmadzadeh, N. Farsad, and
R. Schober, “Constant-composition codes for
maximum likelihood detection without CSI in
diffusive molecular communications,” IEEE Trans.
Commun., vol. 66, no. 5, pp. 1981–1995,
May 2018.

[245] A. Noel and A. W. Eckford, “Asynchronous peak
detection for demodulation in molecular
communication,” in Proc. IEEE Int. Conf.
Commun. (ICC), May 2017, pp. 1–6.

[246] G. Chang, L. Lin, and H. Yan, “Adaptive detection
and ISI mitigation for mobile molecular
communication,” IEEE Trans. Nanobiosci., vol. 17,
no. 1, pp. 21–35, Jan. 2018.

[247] A. Ahmadzadeh, V. Jamali, and R. Schober.
(2017). “Stochastic channel modeling for diffusive
mobile molecular communication systems.”
[Online]. Available:
https://arxiv.org/abs/1709.06785

[248] Y. Murin, N. Farsad, M. Chowdhury, and
A. Goldsmith, “Time-slotted transmission over
molecular timing channels,” Nano Commun.
Netw., vol. 12, pp. 12–24, Jun. 2017.

[249] Y. Murin, N. Farsad, M. Chowdhury, and
A. Goldsmith. (2017). “Optimal detection for
diffusion-based molecular timing channels.”
[Online]. Available:
https://arxiv.org/abs/1709.02977

[250] B. C. Akdeniz, A. E. Pusane, and T. Tugcu,
“Optimal reception delay in diffusion-based
molecular communication,” IEEE Commun. Lett.,
vol. 22, no. 1, pp. 57–60, Jan. 2018.

[251] T. Mora, “Physical limit to concentration sensing
amid spurious ligands,” Phys. Rev. Lett., vol. 115,
no. 3, 2015, Art. no. 038102.

[252] M. Kuscu and O. B. Akan, “Maximum likelihood
detection with ligand receptors for diffusion-based
molecular communications in Internet of bio-nano
things,” IEEE Trans. Nanobiosci., vol. 17, no. 1,
pp. 44–54, Jan. 2018.

Vol. 107, No. 7, July 2019 | PROCEEDINGS OF THE IEEE 1339



Kuscu et al.: Transmitter and Receiver Architectures for MCs

[253] A. Ahmadzadeh, H. Arjmandi, A. Burkovski, and
R. Schober, “Comprehensive reactive receiver
modeling for diffusive molecular communication
systems: Reversible binding, molecule
degradation, and finite number of receptors,” IEEE
Trans. Nanobiosci., vol. 15, no. 7, pp. 713–727,
Oct. 2016.

[254] Y. Deng, A. Noel, M. Elkashlan, A. Nallanathan,
and K. C. Cheung, “Modeling and simulation of
molecular communication systems with a
reversible adsorption receiver,” IEEE Trans. Mol.
Biol. Multi-Scale Commun., vol. 1, no. 4,
pp. 347–362, Dec. 2015.

[255] H. ShahMohammadian, G. G. Messier, and
S. Magierowski, “Modelling the reception process
in diffusion-based molecular communication
channels,” in Proc. IEEE Int. Conf. Commun.
Workshops (ICC), Jun. 2013, pp. 782–786.

[256] C. T. Chou, “Extended master equation models for
molecular communication networks,” IEEE Trans.
Nanobiosci., vol. 12, no. 2, pp. 79–92, Jun. 2013.

[257] G. Aminian, M. F. Ghazani, M. Mirmohseni,
M. Nasiri-Kenari, and F. Fekri, “On the capacity of
point-to-point and multiple-access molecular
communications with ligand-receptors,” IEEE
Trans. Mol. Biol. Multi-Scale Commun., vol. 1,
no. 4, pp. 331–346, Dec. 2016.

[258] M. Kuscu and O. B. Akan, “Modeling
convection-diffusion-reaction systems for
microfluidic molecular communications with
surface-based receivers in Internet of bio-nano
things,” PLoS ONE, vol. 13, no. 2, 2018,
Art. no. e0192202.

[259] M. U. Mahfuz, D. Makrakis, and H. T. Mouftah,
“Strength-based optimum signal detection in
concentration-encoded pulse-transmitted OOK
molecular communication with stochastic
ligand-receptor binding,” Simul. Model. Pract.
Theory, vol. 42, pp. 189–209, Mar. 2014.

[260] C. T. Chou, “A Markovian approach to the optimal
demodulation of diffusion-based molecular
communication networks,” IEEE Trans. Commun.,
vol. 63, no. 10, pp. 3728–3743, Oct. 2015.

[261] C. T. Chou, “Maximum a-posteriori decoding for
diffusion-based molecular communication using
analog filters,” IEEE Trans. Nanotechnol., vol. 14,
no. 6, pp. 1054–1067, Nov. 2015.

[262] G. Muzio, M. Kuscu, and O. B. Akan, “Selective
signal detection with ligand receptors under
interference in molecular communications,” in
Proc. IEEE 19th Int. Workshop Signal Process. Adv.

Wireless Commun. (SPAWC), Jun. 2018,
pp. 1–5.

[263] I. F. Akyildiz, “Nanonetworks: A new frontier in
communications,” in Proc. ACM 18th Annu. Int.
Conf. Mobile Comput. Netw., 2012,
pp. 1–2.

[264] G. Zheng, X. P. A. Gao, and C. M. Lieber,
“Frequency domain detection of biomolecules
using silicon nanowire biosensors,” Nano Lett.,
vol. 10, no. 8, pp. 3179–3183, 2010.

[265] G. S. Kulkarni and Z. Zhong, “Detection beyond
the Debye screening length in a high-frequency
nanoelectronic biosensor,” Nano Lett., vol. 12,
no. 2, pp. 719–723, 2012.

[266] W. Fu, L. Jiang, E. P. van Geest, L. M. C. Lima, and
G. F. Schneider, “Sensing at the surface of
graphene field-effect transistors,” Adv. Mater.,
vol. 29, no. 6, Feb. 2017,
Art. no. 1603610.

[267] B. Atakan and O. B. Akan, “On channel capacity
and error compensation in molecular
communication,” in Transactions on
Computational Systems Biology X. Berlin,
Germany: Springer, 2008, pp. 59–80.

[268] N. Farsad, N. R. Kim, A. W. Eckford, and
C. B. Chae, “Channel and noise models for
nonlinear molecular communication systems,”
IEEE J. Sel. Areas Commun., vol. 32, no. 12,
pp. 2392–2401, Dec. 2014.

[269] S. Abadal and I. F. Akyildiz, “Bio-inspired
synchronization for nanocommunication
networks,” in Proc. IEEE Global Telecommun. Conf.
(GLOBECOM), Dec. 2011, pp. 1–5.

[270] H. ShahMohammadian, G. G. Messier, and
S. Magierowski, “Blind synchronization in
diffusion-based molecular communication
channels,” IEEE Commun. Lett., vol. 17, no. 11,
pp. 2156–2159, Nov. 2013.

[271] L. Lin, C. Yang, M. Ma, S. Ma, and H. Yan, “A clock
synchronization method for molecular
nanomachines in bionanosensor networks,” IEEE
Sensors J., vol. 16, no. 19, pp. 7194–7203,
Oct. 2016.

[272] V. Jamali, A. Ahmadzadeh, and R. Schober,
“Symbol synchronization for diffusive molecular
communication systems,” in Proc. IEEE Int. Conf.
Commun. (ICC), May 2017, pp. 1–7.

[273] M. U. Mahfuz, D. Makrakis, and H. T. Mouftah,
“Concentration-encoded subdiffusive molecular
communication: Theory, channel characteristics,
and optimum signal detection,” IEEE Trans.

Nanobiosci., vol. 15, no. 6, pp. 533–548,
Sep. 2016.

[274] S. M. Mustam, S. K. S. Yusof, and S. Nejatian,
“Multilayer diffusion-based molecular
communication,” Trans. Emerg. Telecommun.
Technol., vol. 28, no. 1, p. e2935, 2017.

[275] V. Jamali, A. Ahmadzadeh, C. Jardin, C. Sticht,
and R. Schober, “Channel estimation for diffusive
molecular communications,” IEEE Trans.
Commun., vol. 64, no. 10, pp. 4238–4252,
Oct. 2016.

[276] A. Noel, K. C. Cheung, and R. Schober, “Joint
channel parameter estimation via diffusive
molecular communication,” IEEE Trans. Mol. Biol.
Multi-Scale Commun., vol. 1, no. 1, pp. 4–17,
Mar. 2015.

[277] V. Singh, M. Tchernookov, and I. Nemenman,
“Effects of receptor correlations on molecular
information transmission,” Phys. Rev. E, Stat. Phys.
Plasmas Fluids Relat. Interdiscip. Top., vol. 94,
no. 2, 2016, Art. no. 022425.

[278] D. Bray, M. D. Levin, and C. J. Morton-Firth,
“Receptor clustering as a cellular mechanism to
control sensitivity,” Nature, vol. 393, no. 6680,
p. 85–88, 1998.

[279] A. Mugler, F. Tostevin, and P. R. ten Wolde,
“Spatial partitioning improves the reliability of
biochemical signaling,” Proc. Nat. Acad. Sci. USA,
vol. 110, pp. 5927–5932, Apr. 2013.

[280] K. Kaizu, W. de Ronde, J. Paijmans, K. Takahashi,
F. Tostevin, and P. R. ten Wolde, “The Berg–Purcell
limit revisited,” Biophys. J., vol. 106, no. 4,
pp. 976–985, 2014.

[281] C. T. Chou, “Molecular circuit for approximate
maximum a posteriori demodulation of
concentration modulated signals,” in Proc. 5th
ACM Int. Conf. Nanosc. Comput. Commun., 2018,
p. 11.

[282] C. T. Chou, “Chemical reaction networks for
maximum likelihood estimation of the
concentration of signalling molecules,” in Proc.
4th ACM Int. Conf. Nanosc. Comput. Commun.,
2017, p. 4.

[283] J.-B. Lalanne and P. François, “Chemodetection in
fluctuating environments: Receptor coupling,
buffering, and antagonism,” Proc. Nat. Acad. Sci.
USA, vol. 112, no. 6, pp. 1898–1903, 2015.

[284] M. Kuscu and O. B. Akan. (2019). “Spectrum
sensing in molecular communications with ligand
receptors.” [Online]. Available:
https://arxiv.org/abs/1901.05540

A B O U T T H E A U T H O R S

Murat Kuscu (Student Member, IEEE)
received the B.Sc. degree in electrical
and electronics engineering from Middle
East Technical University, Ankara, Turkey,
in 2011, and the M.Sc. and Ph.D. degrees
from the Electrical and Electronics Engineer-
ing Department, Koc University, Istanbul,
Turkey, in 2013 and 2017, respectively.
He is currently a Research Assistant with

the Internet of Everything (IoE) Group, Department of Engineering,
University of Cambridge, Cambridge, U.K. His current research
interests include nanoscale and molecular communications, and
the Internet of Everything.

Ergin Dinc (Member, IEEE) received the
B.Sc. degree in electrical and electronics
engineering from Boğaziçi University, Istan-
bul, Turkey, in 2012, and the Ph.D. degree in
electrical and electronics engineering from
Koc University, Istanbul, in 2016.
From 2016 to 2017, he was a Postdoc-

toral Researcher with the KTH Royal Insti-
tute of Technology, Stockholm, Sweden. He
is currently a Postdoctoral Research Associate with the Electrical
Engineering Division, University of Cambridge, Cambridge, U.K. He
conducts theoretical and experimental research on molecular com-
munication, neural communication, and cyber–physical systems.

1340 PROCEEDINGS OF THE IEEE | Vol. 107, No. 7, July 2019



Kuscu et al.: Transmitter and Receiver Architectures for MCs

Bilgesu A. Bilgin (Member, IEEE) received
the B.S. degree in electrical and electron-
ics engineering from Middle East Techni-
cal University, Ankara, Turkey, in 2008, and
the M.Sc. and Ph.D. degrees in mathemat-
ics from Koc University, Istanbul, Turkey,
in 2011 and 2015, respectively.
From 2015 to 2017, he was a Postdoctoral

Fellow with the Next-generation and Wire-
less Communications Laboratory, Koc University. He is currently a
Postdoctoral Research Associate with the Engineering Department,
University of Cambridge, Cambridge, U.K. His current research
interests include molecular communication, intrabody nanonet-
works, partial differential equations, and dynamical systems.

Hamideh Ramezani (Student Member,
IEEE) received the B.S. degree in electri-
cal engineering, major in communication,
from the University of Tehran, Tehran, Iran,
in 2011, the M.S. degree in electrical engi-
neering, major in communication systems,
from the Sharif University of Technology,
Tehran, in 2013, and the Ph.D. degree in
electrical and electronics engineering from
Koc University, Istanbul, Turkey, in 2018. She is currently working
toward the Ph.D. degree at the Internet of Everything (IoE) Group,
Department of Engineering, University of Cambridge, Cambridge,
U.K.
She is a Research Assistant with the Internet of Everything (IoE)

Group, Department of Engineering, University of Cambridge. Her
current research interests include biosensing, nanoscale neural and
molecular communications, and the Internet of Everything.

Ozgur B. Akan (Fellow, IEEE) received the
Ph.D. degree from the Broadband and Wire-
less Networking Laboratory, School of Elec-
trical and Computer Engineering, Georgia
Institute of Technology, Atlanta, GA, USA,
in 2004.
He is currently the Head of the Electri-

cal Engineering Division, Internet of Every-
thing (IoE) Group, Department of Engineer-
ing, University of Cambridge, Cambridge, U.K., and the Director
of the Next-Generation and Wireless Communications Laboratory,
Department of Electrical and Electronics Engineering, Koc Univer-
sity, Istanbul, Turkey. His current research interests include wire-
less, nano, molecular, and neural communications, and the Internet
of Everything.
Dr. Akan is a Series Editor of the IEEE Communications Magazine,

an Inaugural Associate Editor of the IEEE NETWORKING LETTERS,
and an Associate Editor of the IET Communications and the Nano
Communication Networks journal.

Vol. 107, No. 7, July 2019 | PROCEEDINGS OF THE IEEE 1341



<<
  /ASCII85EncodePages false
  /AllowTransparency false
  /AutoPositionEPSFiles true
  /AutoRotatePages /None
  /Binding /Left
  /CalGrayProfile (Gray Gamma 2.2)
  /CalRGBProfile (sRGB IEC61966-2.1)
  /CalCMYKProfile (U.S. Web Coated \050SWOP\051 v2)
  /sRGBProfile (sRGB IEC61966-2.1)
  /CannotEmbedFontPolicy /Warning
  /CompatibilityLevel 1.4
  /CompressObjects /Off
  /CompressPages true
  /ConvertImagesToIndexed true
  /PassThroughJPEGImages true
  /CreateJobTicket false
  /DefaultRenderingIntent /Default
  /DetectBlends true
  /DetectCurves 0.0000
  /ColorConversionStrategy /sRGB
  /DoThumbnails true
  /EmbedAllFonts true
  /EmbedOpenType false
  /ParseICCProfilesInComments true
  /EmbedJobOptions true
  /DSCReportingLevel 0
  /EmitDSCWarnings false
  /EndPage -1
  /ImageMemory 1048576
  /LockDistillerParams true
  /MaxSubsetPct 100
  /Optimize true
  /OPM 0
  /ParseDSCComments false
  /ParseDSCCommentsForDocInfo true
  /PreserveCopyPage true
  /PreserveDICMYKValues true
  /PreserveEPSInfo false
  /PreserveFlatness true
  /PreserveHalftoneInfo true
  /PreserveOPIComments false
  /PreserveOverprintSettings true
  /StartPage 1
  /SubsetFonts false
  /TransferFunctionInfo /Remove
  /UCRandBGInfo /Preserve
  /UsePrologue false
  /ColorSettingsFile ()
  /AlwaysEmbed [ true
    /Arial-Black
    /Arial-BoldItalicMT
    /Arial-BoldMT
    /Arial-ItalicMT
    /ArialMT
    /ArialNarrow
    /ArialNarrow-Bold
    /ArialNarrow-BoldItalic
    /ArialNarrow-Italic
    /ArialUnicodeMS
    /BookAntiqua
    /BookAntiqua-Bold
    /BookAntiqua-BoldItalic
    /BookAntiqua-Italic
    /BookmanOldStyle
    /BookmanOldStyle-Bold
    /BookmanOldStyle-BoldItalic
    /BookmanOldStyle-Italic
    /BookshelfSymbolSeven
    /Century
    /CenturyGothic
    /CenturyGothic-Bold
    /CenturyGothic-BoldItalic
    /CenturyGothic-Italic
    /CenturySchoolbook
    /CenturySchoolbook-Bold
    /CenturySchoolbook-BoldItalic
    /CenturySchoolbook-Italic
    /ComicSansMS
    /ComicSansMS-Bold
    /CourierNewPS-BoldItalicMT
    /CourierNewPS-BoldMT
    /CourierNewPS-ItalicMT
    /CourierNewPSMT
    /EstrangeloEdessa
    /FranklinGothic-Medium
    /FranklinGothic-MediumItalic
    /Garamond
    /Garamond-Bold
    /Garamond-Italic
    /Gautami
    /Georgia
    /Georgia-Bold
    /Georgia-BoldItalic
    /Georgia-Italic
    /Haettenschweiler
    /Impact
    /Kartika
    /Latha
    /LetterGothicMT
    /LetterGothicMT-Bold
    /LetterGothicMT-BoldOblique
    /LetterGothicMT-Oblique
    /LucidaConsole
    /LucidaSans
    /LucidaSans-Demi
    /LucidaSans-DemiItalic
    /LucidaSans-Italic
    /LucidaSansUnicode
    /Mangal-Regular
    /MicrosoftSansSerif
    /MonotypeCorsiva
    /MSReferenceSansSerif
    /MSReferenceSpecialty
    /MVBoli
    /PalatinoLinotype-Bold
    /PalatinoLinotype-BoldItalic
    /PalatinoLinotype-Italic
    /PalatinoLinotype-Roman
    /Raavi
    /Shruti
    /Sylfaen
    /SymbolMT
    /Tahoma
    /Tahoma-Bold
    /TimesNewRomanMT-ExtraBold
    /TimesNewRomanPS-BoldItalicMT
    /TimesNewRomanPS-BoldMT
    /TimesNewRomanPS-ItalicMT
    /TimesNewRomanPSMT
    /Trebuchet-BoldItalic
    /TrebuchetMS
    /TrebuchetMS-Bold
    /TrebuchetMS-Italic
    /Tunga-Regular
    /Verdana
    /Verdana-Bold
    /Verdana-BoldItalic
    /Verdana-Italic
    /Vrinda
    /Webdings
    /Wingdings2
    /Wingdings3
    /Wingdings-Regular
    /ZWAdobeF
  ]
  /NeverEmbed [ true
  ]
  /AntiAliasColorImages false
  /CropColorImages true
  /ColorImageMinResolution 200
  /ColorImageMinResolutionPolicy /OK
  /DownsampleColorImages true
  /ColorImageDownsampleType /Bicubic
  /ColorImageResolution 600
  /ColorImageDepth -1
  /ColorImageMinDownsampleDepth 1
  /ColorImageDownsampleThreshold 1.50000
  /EncodeColorImages true
  /ColorImageFilter /DCTEncode
  /AutoFilterColorImages false
  /ColorImageAutoFilterStrategy /JPEG
  /ColorACSImageDict <<
    /QFactor 0.15
    /HSamples [1 1 1 1] /VSamples [1 1 1 1]
  >>
  /ColorImageDict <<
    /QFactor 0.76
    /HSamples [2 1 1 2] /VSamples [2 1 1 2]
  >>
  /JPEG2000ColorACSImageDict <<
    /TileWidth 256
    /TileHeight 256
    /Quality 30
  >>
  /JPEG2000ColorImageDict <<
    /TileWidth 256
    /TileHeight 256
    /Quality 30
  >>
  /AntiAliasGrayImages false
  /CropGrayImages true
  /GrayImageMinResolution 200
  /GrayImageMinResolutionPolicy /OK
  /DownsampleGrayImages true
  /GrayImageDownsampleType /Bicubic
  /GrayImageResolution 600
  /GrayImageDepth -1
  /GrayImageMinDownsampleDepth 2
  /GrayImageDownsampleThreshold 1.50000
  /EncodeGrayImages true
  /GrayImageFilter /DCTEncode
  /AutoFilterGrayImages false
  /GrayImageAutoFilterStrategy /JPEG
  /GrayACSImageDict <<
    /QFactor 0.15
    /HSamples [1 1 1 1] /VSamples [1 1 1 1]
  >>
  /GrayImageDict <<
    /QFactor 0.76
    /HSamples [2 1 1 2] /VSamples [2 1 1 2]
  >>
  /JPEG2000GrayACSImageDict <<
    /TileWidth 256
    /TileHeight 256
    /Quality 30
  >>
  /JPEG2000GrayImageDict <<
    /TileWidth 256
    /TileHeight 256
    /Quality 30
  >>
  /AntiAliasMonoImages false
  /CropMonoImages true
  /MonoImageMinResolution 400
  /MonoImageMinResolutionPolicy /OK
  /DownsampleMonoImages true
  /MonoImageDownsampleType /Bicubic
  /MonoImageResolution 1200
  /MonoImageDepth -1
  /MonoImageDownsampleThreshold 1.50000
  /EncodeMonoImages true
  /MonoImageFilter /CCITTFaxEncode
  /MonoImageDict <<
    /K -1
  >>
  /AllowPSXObjects false
  /CheckCompliance [
    /None
  ]
  /PDFX1aCheck false
  /PDFX3Check false
  /PDFXCompliantPDFOnly false
  /PDFXNoTrimBoxError true
  /PDFXTrimBoxToMediaBoxOffset [
    0.00000
    0.00000
    0.00000
    0.00000
  ]
  /PDFXSetBleedBoxToMediaBox true
  /PDFXBleedBoxToTrimBoxOffset [
    0.00000
    0.00000
    0.00000
    0.00000
  ]
  /PDFXOutputIntentProfile (None)
  /PDFXOutputConditionIdentifier ()
  /PDFXOutputCondition ()
  /PDFXRegistryName ()
  /PDFXTrapped /False

  /CreateJDFFile false
  /Description <<
    /CHS <FEFF4f7f75288fd94e9b8bbe5b9a521b5efa7684002000410064006f006200650020005000440046002065876863900275284e8e55464e1a65876863768467e5770b548c62535370300260a853ef4ee54f7f75280020004100630072006f0062006100740020548c002000410064006f00620065002000520065006100640065007200200035002e003000204ee553ca66f49ad87248672c676562535f00521b5efa768400200050004400460020658768633002>
    /CHT <FEFF4f7f752890194e9b8a2d7f6e5efa7acb7684002000410064006f006200650020005000440046002065874ef69069752865bc666e901a554652d965874ef6768467e5770b548c52175370300260a853ef4ee54f7f75280020004100630072006f0062006100740020548c002000410064006f00620065002000520065006100640065007200200035002e003000204ee553ca66f49ad87248672c4f86958b555f5df25efa7acb76840020005000440046002065874ef63002>
    /DAN <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>
    /DEU <FEFF00560065007200770065006e00640065006e0020005300690065002000640069006500730065002000450069006e007300740065006c006c0075006e00670065006e0020007a0075006d002000450072007300740065006c006c0065006e00200076006f006e002000410064006f006200650020005000440046002d0044006f006b0075006d0065006e00740065006e002c00200075006d002000650069006e00650020007a0075007600650072006c00e40073007300690067006500200041006e007a006500690067006500200075006e00640020004100750073006700610062006500200076006f006e00200047006500730063006800e40066007400730064006f006b0075006d0065006e00740065006e0020007a0075002000650072007a00690065006c0065006e002e00200044006900650020005000440046002d0044006f006b0075006d0065006e007400650020006b00f6006e006e0065006e0020006d006900740020004100630072006f00620061007400200075006e0064002000520065006100640065007200200035002e003000200075006e00640020006800f600680065007200200067006500f600660066006e00650074002000770065007200640065006e002e>
    /ESP <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>
    /FRA <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>
    /ITA (Utilizzare queste impostazioni per creare documenti Adobe PDF adatti per visualizzare e stampare documenti aziendali in modo affidabile. I documenti PDF creati possono essere aperti con Acrobat e Adobe Reader 5.0 e versioni successive.)
    /JPN <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>
    /KOR <FEFFc7740020c124c815c7440020c0acc6a9d558c5ec0020be44c988b2c8c2a40020bb38c11cb97c0020c548c815c801c73cb85c0020bcf4ace00020c778c1c4d558b2940020b3700020ac00c7a50020c801d569d55c002000410064006f0062006500200050004400460020bb38c11cb97c0020c791c131d569b2c8b2e4002e0020c774b807ac8c0020c791c131b41c00200050004400460020bb38c11cb2940020004100630072006f0062006100740020bc0f002000410064006f00620065002000520065006100640065007200200035002e00300020c774c0c1c5d0c11c0020c5f40020c2180020c788c2b5b2c8b2e4002e>
    /NLD (Gebruik deze instellingen om Adobe PDF-documenten te maken waarmee zakelijke documenten betrouwbaar kunnen worden weergegeven en afgedrukt. De gemaakte PDF-documenten kunnen worden geopend met Acrobat en Adobe Reader 5.0 en hoger.)
    /NOR <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>
    /PTB <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>
    /SUO <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>
    /SVE <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>
    /ENU (Use these settings to create PDFs that match the "Required"  settings for PDF Specification 4.0)
  >>
>> setdistillerparams
<<
  /HWResolution [300 300]
  /PageSize [576.000 782.640]
>> setpagedevice