Design and Evaluation of a Receiver for Wired Nano-Communication Networks
11 Design and Evaluation of a Receiver for WiredNano-Communication Networks
Oussama Abderrahmane Dambri , Student Member, IEEE and Soumaya Cherkaoui , Senior Member, IEEE
Abstract —In this paper, we propose a bio-inspired receiver,which detects the electrons transmitted through a nanowire,then, it converts the detected information into a blue light usingbioluminescence. Using light allows the designed receiver to alsoact as a relay for the nearest gateway (photo-detector). Wesimulate the construction of the nanowire, present its electricalcharacteristics and calculate its maximum throughput for a betterdesign of the receiver. The designed receiver contains two parts,a part that detects the transmitted electrons, which we model byusing an equivalent circuit, and a part that converts the detectedelectrons into a blue light. We derive the analytical expressions ofthe equivalent circuit’s components, and we calculate the emittedphotons for each electrical pulse detected. We also proposemodulation techniques that guaranty an effective decoding ofthe information. We send a binary message and we follow theelectron detection process of the proposed receiver until lightemission and we calculate the Bit Error Rate (BER) to evaluatethe performance of the designed receiver. The results of thisstudy show that the designed receiver can accurately detect theelectrons sent through a conductive nanowire in wired nano-communication networks, and that it can also act as a relay forthe nearest gateway.
Index Terms —Wired, Nano-Communication, Bioluminescence,Receiver, Aequorin, Calcium, Throughput, Modulation.
I. I
NTRODUCTION N ANOTECHNOLOGY has become a key area of researchin multidisciplinary fields, and its rapid and impressiveadvance has led to new applications in biomedical and militaryindustries. It turned out that nanomachines need a cooperativebehavior to overcome their limited-processing capacities inorder to achieve a common objective. Exploiting the poten-tial advantages of nano-communication networks between thenanomachines drove the researchers to come up with newsolutions to create systems that communicate at nanoscale.Three methods are proposed in literature, wireless electromag-netic method using THz band [1]–[4], bio-inspired wirelessmolecular communication using molecules [5]–[13] and wirednano-communication method using polymers [14].However, on the one hand, THz band suffers from a veryhigh path loss, mainly caused by water molecules absorption,which complicates its use in medical applications inside thehuman body [2]–[4]. On the other hand, molecular communi-cation biocompatibility makes it very promising for medicalapplications at nanoscale, such as monitoring, diagnosis andlocal drug delivery [5]. Nevertheless, the achievable through-put of molecular communication systems is very limitedand the delay is very high due to the random distributionof molecules in the medium [6]–[9]. Moreover, molecularcommunication systems suffer from intersymbol interference, despite the proposed techniques in literature to reduce it [10]–[13], [15].Wired nano-communication is a new proposed method thatuses the self-assembly ability of some polymers inside livingcells to build an electrically conductive nanowire [14]. Thismethod has the biocompatibility of molecular communication,but also, it has a very high achievable throughput sincefast electrons are the carriers of information. One of themain challenges of the wired nano-communication systemis detecting the electrons at the receiver without losing itsbiocompatibility. The proposed receivers in the literature formolecular communication are designed to observe or absorbmolecules in the medium [16]–[19], and thus, they cannot beused to detect electrons for wired nano-communication.In our recent works [14], [20], we proposed the idea ofa receiver for a wired nano-communication system withoutstudying or evaluating it. In this paper, we present the detailedsystem model of the proposed receiver, and we evaluateits performance. The proposed receiver contains a SmoothEndoplasmic Reticulum (SER), which plays the role of Ca storage inside living cells [21]. The receiver also contains arecyclable concentration of a photo-protein Aequorin , whichcan be found in the jellyfish
Aequorea Victoria that livesin North America and the Pacific Ocean [22]. The electrondetection at nanoscale is extremely difficult due to the quantumtrade-off between information and uncertainty. Therefore, thedesigned receiver uses the transmitted electrons to stimulateSER and trigger a chemical reaction that generates a bio-luminescent light by using the photo-protein Aequorin, asshown in Fig. 1. In order to study SER stimulation with thetransmitted electrons, we simulated the conductive nanowireused in wired nano-communication systems, we presentedits electrical characteristics and we calculated its maximumthroughput. Bioluminescence is a chemical reaction used byliving organisms to generate light with enzymes and photo-proteins. Converting the transmitted electron pulses into lightpulses at the receiver makes the extraction of informationeasier and can create a link between nanoscale and macroscalecommunication systems.The rest of the paper is organized as follows. In sectionII, we present and explain the 2 algorithms used in ourframework to simulate the nanowires assembly in a stochastic3D system. We also present the electrical characteristics of thenanowire and we calculate its maximum throughput and theerror probability. Section III highlights an in-depth descriptionof the designed receiver, by clarifying the SER role in electrondetection, and by explaining the triggered chemical reactionthat emits a blue light. In section IV, we model the designed a r X i v : . [ c s . ET ] S e p receiver by using an equivalent circuit, where SER and itsmembrane represent a capacitor and the sum of Ca+2 channelsrepresent a resistance linked in parallel. We derive the analyt-ical expression of the circuit components and we calculatethe emitted photons for each electron pulse detected. We alsocalculate the Bit Error Rate (BER) of the designed receiver. Insection V, we propose modulation techniques that guaranty aneffective decoding of the detected information at the designedreceiver. Section VI discusses the numerical results of thisstudy, by following the detection process of a random binarymessage at the receiver until its conversion into a blue light. Italso presents the results of BER to evaluate the performanceof the designed receiver. We conclude the paper in section VII.II. S IMULATION F RAMEWORK
GlowScript VPython is an easy and efficient way to create3D environments for real time simulations. Our frameworkapplies GlowScript as a browser-based implementation thatruns a VPython program by using RapydScript, which is aPython-to-JavaScript compiler. In this study, we simulate theself-assembly of a polymer called ”actin” to create a nanowirein real time, by using sphere molecules that diffuse randomlyin a 3D cube as shown in Fig. 2. By using this framework, wecan follow the position of the last assembled molecule in realtime to study the speed of the nanowire’s construction as thegraph in Fig. 3 exhibits. In the next subsections, we presentand explain the two algorithms that we used in our framework.
A. Collision Between Actin Molecules
The algorithm 1 is a combination of two strategies, PeriodicInterference Test (PIT) and a Predicted Instant of Collision(PIC). The (PIT) consists of checking to detect if any collisionhas occurred between two molecules, at every frame in thesimulation. It tests if the molecules are approaching eachother, and whether they interfere spatially. When a collisionis detected, the algorithm computes the new velocities afterthe shock using the momentum and energy conservationequations; and then attributes the new velocity to the movingmolecules after the collision. If the simulation detects nocollision in that frame, it continues to the next one. ThePIC, on the other hand, pre-calculates the exact time ofcollisions before spatial interactions between the molecules.The difference between PIT and PIC approaches is that thelater perfectly models the collisions, at the cost of some extracomputation, compared to the PIT approach, which is a lazycollision detection strategy. In our algorithm 1, we used acombination of the two approaches to perfectly detect andmodel the collisions in real time.
B. Nanowire Formation
If a random molecule in the environment strikes the lastmolecule of the constructed nanowire, it will attach to itincreasing the length of the nanowire. Several conditions arenecessary for the molecules to stick in the nanowire, namely: • One of the molecules having a collision is already stick-ing in the nanowire.
Fig. 1. The designed receiver for wired nano-communication networks.
Algorithm 1
Periodic Interference Test procedure PITC( M
OLECULE
A, M
OLECULE
B ): At every frame instant t: Test if the two molecules are approaching each other. if YES then Test if they overlap (interfere) spatially. if YES then Collision detected. Compute New Velocities. T:= time of collision between A and B. if T < next frame instant: then Move frame to intermediate instant T.
Calculate velocities after collision. end if else
No collision detected. end if
Goto next frame instant. else : Goto next frame instant. end if end procedure • The center of the striking molecule lies within a specifiedangle range from the center of the last attached molecule. • The center of the next molecule being attached shouldbe farther away from the transmitter and closer to thereceiver, so that the nanowire is moving towards thereceiver and not towards any random path.To make the molecules stick, we give their momentum azero value, and we do not update their position with time.The algorithm 2 also regulates the direction of the nanowireformation. The direction is controlled by the magnetic field.If the magnetic field is zero, then there is no guide for thedirection, and the nanowire forms randomly. As the magneticfield increases, the nanowire gets more and more aligned tothe straight line joining the transmitter and the receiver, asconfirmed by the experimental results in [23].
Fig. 2. Actin-based nanowire formation represented by small black spheres,which link the transmitter to the receiver (cyan spheres).
Algorithm 2
Nanowire Formation molecule array = [Transmitter] if collision.detected = True then if collision occurs with last(molecule array) &&last(molecule array) != Receiver then M ← molecule hitting the wire if Magnetic Field = 0 then Randomly attach M to last(molecule array) molecule array.append(M) end if if Magnetic Field != 0 then if M.center.X > max(last(molecule array) .po-sition) then Z = function(magnetic field) if M.center.Z < max(Z) then M.momentum = 0 % Stick the twomolecules together molecule array.append(M) end if end if end if end if end if
C. Channel’s Electrical Characteristics
The authors in [24] and [25] studied the electrical impulsesand ionic waves propagating along actin filaments in bothintracellular and in vitro conditions, by modeling the actinfilament as an RLC equivalent circuit. The effective resistance,inductance and capacitance for a 1 µm actin filament are [25]: C eq = 0.02 × − F L eq = 340 × − H R eq = 1.2 × Ω The Fig. 5 represents actin nanowire’s electrical characteris-
Fig. 3. Position of the last assembled molecule as a function of timerepresenting the speed of the nanowire formation.Fig. 4. Maximum throughput approximation of the actin nanowire. tics simulated in our recent work [26], in terms of attenuation,phase and delay as a function of the frequency and the distancebetween the transmitter and the receiver. We can see thatthe attenuation of actin filaments is very high because ofits resistance. The experimental study in [24] showed thatdespite the actin filaments attenuation, the velocity of thecharges passing through it can reach 30,000 µm / s and thenslows down rapidly in the first 60 µs . Moreover, the authorsin [27] proved experimentally that adding gold nanoparticlesto the actin monomers drastically decreases its attenuation. D. Channel’s Maximum Throughput
The charge capacity of each actin monomer is ∼ e [25],and by assuming 370 monomers/ µs of an actin filament, wecan deduce that the charge capacity of an actin filament with1 µm . long is approximately 1480 e . By assuming that 1 e represents 1 bit of information, the maximum throughput iscalculated by multiplying the total charge capacity of 1 µs actin nanowire by the speed of the charges propagation throughit and we write [26]: T M ( t ) = v ( t ) × ψ tot . (1)where T M ( t ) represents the nanowire achievable throughput, v ( t ) is the charge’s propagation speed calculated in [24] and ψ tot is the total charge capacity of 1 µs actin nanowire.Fig. 4 shows the maximum throughput approximation of theproposed nanowire. The decrease of the achievable throughput Fig. 5. The electrical characteristics of actin nanowires in terms of attenuation, phase and delay as a function of the frequency and the channel distance. over time is due to the high attenuation of the actin filaments.However, despite this decrease, and to the best of our knowl-edge, the throughout is still very high compared to molecularcommunication results proposed in the literature so far.
E. Error Probability
The error probability of the proposed polymer-based chan-nel for wired nano-communication systems can be written as: P e = p P (cid:15) + p P (cid:15) , (2)Where p = p = 0.5 are the a priori probabilities, P (cid:15) is theprobability of error for a bit ’0’ and P (cid:15) is the probabilityof error for a bit ’1’. The constant assembly and disassemblyof actin monomers to construct the actin nanowire generates anoise, which we modeled as a normal distribution. The error ismore probable in bit ’0’ than in bit ’1’, because the assemblyof the nanowire is more frequent than the disassembly. So,the distribution of bit ’1’ is skewed left compared to thedistribution of bit ’0’, which remains Gaussian and we write[14]: P (cid:15) = 1 √ π exp (cid:18) − x (cid:19) , (3) P (cid:15) = 1 √ π exp (cid:18) − ( x − A ) (cid:19) erfc (cid:18) x − A √ (cid:19) , (4)Where A is the skewness coefficient of the distribution. Bysubstituting (3) and (4) in (2), we write the probability as: P e = 12 √ π (cid:20) exp (cid:18) − ( x − A ) (cid:19) erfc (cid:18) x − A √ (cid:19) + exp (cid:18) − x (cid:19)(cid:21) . (5)Since the distribution is negatively skewed, A is also nega-tive. For a skew normal distribution for which the scale factoris 1, the variance is given by − δ π , where δ = A √ A .III. R ECEIVER D ESIGN
From chlorophyll pigments in plants to the neural system inhuman brains, biological systems always use chemical meansto detect and send electrons in the medium. Inspired bymuscle fiber contraction and relaxation process, we proposein this paper a receiver design for wired nano-communicationnetworks, which uses SER to detect transmitted electronsthrough a nanowire. To avoid the absorption of electrons bythe receiver’s surface, it can be constructed with a hybrid phospholipid/alkanethiol bilayers membrane proposed in [28],because of its insulating nature, in order to minimize theloss of electrons at the receiver. The nucleation of monomersthat trigger the formation of the actin self-assembly can betethered at the bio-engineered membrane of the receiver byusing a polyethylene glycol (PEG) on one end and an electrodeon the other end [29]. The insertion of the monomer canbe spontaneous, electrochemical or with a proteoliposomeinsertion as explained in [29] with details. When the assemblednanowire reaches the receiver, it binds to one of the monomersalready anchored to the receiver’s membrane with electrodes,which creates a passage of electrons through the insulatingmembrane. As shown in Fig. 1, the designed receiver containsSER that stocks Ca ions and a photo-protein that emits ablue light in the presence of Ca ions. The emitted light isdetected by a photo-detector (gateway). A. Electrons Detection
The Ca ion distribution is used in one of the mostimportant biological signaling between living cells, amongother functions such as hormone regulation, muscles con-traction and neurons excitation [30]. The concentration ofCa ions inside the cells must be regulated all the timeusing a very complex system, where SER plays the role ofCa ions storage. The smooth endoplasmic reticulum, alsocalled sarcoplasmic reticulum in muscle cells, is a tubularstructure organelle found in most living cells. The capacityof SER at stocking Ca ions is huge because of a buffercalled calsequestrin that can bind to around 50 Ca , whichdecreases the amount of free Ca inside SER, and therefore,more calcium can be stored [31]. When SER is stimulatedwith an electrical stimulus, the calcium channels open andCa ions get released so fast inside the cells as shown inFig. 6. The Ca concentration is so small inside the cellsthat a tiny increase in their concentration is detected.The SER inside the designed receiver is linked to thenanowire with an electrode so that the transmitted electronscan stimulate the SER, which increases its membrane voltageand open calcium channels. The number of opened calciumchannels, and thus, the released concentration of Ca is pro-portional to the intensity of the electrical current stimulatingthe SER. A small electrical current in the order of picoampere(pA) is sufficient to stimulate SER and release a micromole( µ M) of Ca [32]. When the electrical current stops, calciumchannels close and the Ca ions will be absorbed and stocked again inside the SER. The short-time presence of Ca ions inside the receiver triggers the bioluminescent chemicalreaction, which emits a blue light. B. Light Emission
Photo-proteins are priceless biochemical tools for a varietyof fields including drug discovery, protein dynamic studies andgene expression analysis.
Aequorin is a very important photo-protein for biological studies because it helps researchersmeasure and study the Ca distribution in vivo. Upon bindingto Ca ions, the oxidation of Aequorin molecule is triggered,resulting in the emission of a bioluminescent blue light (470nm). There are other photo-proteins in nature with differentwavelengths emissions such as
Luciferin (530 nm) and somechromophores (630 nm). However, unlike other biolumines-cent reactions that involve the oxidation of an organic substratesuch as
Luciferin and chromophores, adding molecular oxygenis not required in Ca -dependent light emissions, because Ae-quorin protein has oxygen bound to it [33]. Other advantagesof using
Aequorin are that it does not involve any diffusibleorganic factor, no direct participation of enzymes and that itcan be recycled after use [34].
Aequorin is extracted from thejellyfish
Aequorea Victoria that lives in North America and thePacific Ocean [22], and then purified in distilled water. Thisphoto-protein is very sensitive to changes in the concentrationof free Ca , which explains its extensive use as an indicatorof Ca ions in biological studies. In the presence of a verylow concentration of Ca , the light intensity emitted by Aequorin will be independent from Ca concentration, thebioluminescent reaction becomes Ca -dependent only whenthe concentration exceeds 10 − M.The designed receiver uses the short-time presence of thereleased Ca ions to trigger the oxidation of Aequorin, whichemits a photon for each 3 Ca ions bound, as shown in Fig.7. This mechanism of emitting-light can trigger the release ofover 70 kcal of energy as a visible radiation through a singletransition [34]. In the next section, we model the electrondetector as an RC circuit, and we calculate the radiant energyemitted from the bioluminescent reaction at each symbolinterval. IV. R ECEIVER M ODEL
Cell membranes are biological structures that surround cellsand separate their interior from the external environment forthe protection and control of ion exchanges. The membraneis constructed with phospholipid molecules, where the lipidend is hydrophobic and the phosphate end is hydrophilic.When lipid ends get together to form a double-layered sheetand close the sphere, they create a spherical surface thatperfectly separates two volumes of liquid. However, a purephospholipid bilayer is an excellent insulator which does notallow any ion exchange, thus, other pores penetrating themembrane are necessary, allowing the ions to circulate insideand outside the cell. These pores are called ion channels andby controlling their opening and closing, the cell controls theions flow. Cell membranes have specific channels for eachion type, and closing them creates a difference between the
Fig. 6. Ca ions release by a Smooth Endoplasmic Reticulum (SER).Fig. 7. Aequorin bioluminescent reaction that generates blue light in thepresence of Ca ions. ion’s concentration inside and outside the cell, which set up aspecific potential for each ion type.Let’s consider V i the reversal potential of ion species i ,which is the value of the membrane potential for whichthe flux of ion species i is zero, and let’s consider R i thechannel resistance, which is simply the inverse of channel’sconductance. According to Ohm’s law, the ions flow acrossthe channel is proportional to the reversal potential, and theproportionality factor is the channel conductance g i and wecan write: I i = V i R i = g i V i , (6)where I i is the current of ion species i that flows acrossthe membrane. The current flows until reaches an equilibriumcalled the resting potential V , which can be calculated as: V = (cid:80) i g i V i (cid:80) i g i , (7)In our case, the membrane has only specific channels forCa ions, therefore, g i becomes g Ca and V i becomes V Ca .The reversal potential V Ca is calculated by using Nernstequation as: V Ca = kTze ln P out P in , (8)where k is the Boltzmann constant, e is the electron charge, T is the temperature in Kelvin, z is the valence of the Ca ion, P out and P in are the probabilities of finding a Ca ion outsideor inside the SER respectively. When the membrane is at rest, V Ca equals V , however, when an external potential is applied, V Ca increases which opens the Ca channels and dischargesSER from Ca ions. When the excitation is passed, specialpores called pumps charges SER again with Ca ions. Thecharge and discharge of SER can be modeled as an equivalentcapacitance and the inverse of the channel conductance can bemodeled as an equivalent resistance in series with a voltagesource. Therefore, the SER and its membrane can be modeledas an equivalent RC circuit as shown in Fig. 8. A. The Capacitance
The SER membrane separates two conductive liquids thatcontain free ions, so we have two conductors separated by aninsulator. The potential difference across the membrane thatseparates a charge of a Ca ion Q Ca defines a capacitance C Ca that can be written as: C Ca = Q Ca | ∆ V Ca | , (9)Before calculating the potential difference ∆ V Ca by usingthe Gauss’s law, we define the permittivity ε of the membraneas ε = ε ε r , where ε r is the relative permittivity. SER inmuscle cells is composed of tubule networks called cisternae,which they have a diameter r SER = 50 nm . The SER tubulesextend throughout muscle fiber filaments, so we assume thatthe length of SER tubules used in our designed receiver is l = 1 µm . By assuming that the enclosed charge within SER’svolume is Q Ca , then from Gauss’s law we can write theelectric field at a distance r as: (cid:126)E Ca = Q Ca ε (cid:18) πlr (cid:19) ˆ r. (10)where ˆ r is the radial vector from the Ca charge at theorigin of SER tubules to the surface of the membrane. Thepotential difference between the inside and the outside of SERmembrane is therefore written as: ∆ V Ca = − (cid:90) r SER + δr SER E Ca .dr = − Q Ca πlr ln (cid:18) δr SER (cid:19) , (11)where δ is the thickness of the membrane. By substituting eq.6 in eq. 4 we find: C Ca = 2 πlr ln (cid:0) δr SER (cid:1) . (12)The membrane of SER does not need a high voltageto separate charges because it is only two molecules thick( δ = 6 × − m ), thus, we expect the membrane capacitanceto be quite high per unit area. With the parameters given above,we estimate that the capacitance is C Ca (cid:39) × − pF/µm .Unlike the conductance, the capacitance of biological mem-branes is not influenced by the complexities of biologicalsystems, which makes it constant. Fig. 8. The proposed equivalent RC circuit of SER’s membrane.
B. The Resistance
The pure phospholipid bilayer constructing the membrane isan excellent electrical insulator with a very low conductance,but the mosaic proteins that span the surface of the membraneact as channels for ions and increase the conductance. TheCa current flowing through SER membrane depends onthe number of open channels by the applied external voltage.Depending on the type and the age of SER, the number ofchannels varies between 3.8 and 24.3 per µm area [35].By taking into account the infinitesimal area of channels, weconsider the SER membrane as an infinite charged linear lineand by using Gauss’s law we can write the Ca current I Ca of one channel as: I Ca = σ (cid:90) E Ca .da = σ λlε , (13)where σ is the conductivity of one Ca channel and λ = Q Ca /l . From eq. 1 we can calculate the resistance of Ca channels R Ca : R Ca = V Ca I Ca , (14)and by substituting eq. 6 and eq. 8 in eq. 9 we find: R Ca = ln (cid:0) δr SER (cid:1) πnσl . (15)where n is the number of Ca channels. R Ca is the sumof n variable resistances linked in parallel in our equivalentcircuit. The conductivity of a Ca channel is estimated tobe σ = 0 . pS [36], and thus, the mean value of the variableresistance in eq. 9 can be approximated to R Ca (cid:39) . M Ω .The membrane resistance is highly variable because it is sothin that a very small change in the voltage can create a strongelectric field within it. C. Radiant Energy
The presence of Ca ions near the photo-protein Aequorintriggers a bioluminescent reaction that emits a blue light. Weassume that Ca ions and Aequorin molecules have homoge-nous distributions inside the receiver and that for each 3 Ca ions released outside SER, there is an Aequorin molecule thatbinds to them and emits one photon. Radiant energy is anelectromagnetic energy carried by a stream of photons, thus,the sum of photons emitted by the bioluminescent reactionrepresents the radiant energy of the designed receiver. With theassumptions we made above, we calculate the radiant energyof the designed receiver for each symbol interval as follows: Q e = 13 hcλ (cid:90) T p.c ( t ) dt, (16)where h is the Planck constant, c is the speed of light, λ isthe photon wavelength. T is the symbol time, and p is theprobability for each Aequorin molecule to emit a photon. Inthis paper, the concentration of the released Ca ions for eachsymbol interval c ( t ) is obtained numerically by using Simulinkin MATLAB, which represents the output of the capacitor inthe proposed equivalent circuit. An analytical expression forthe released Ca ion concentration will be derived in ourfuture work. V. M ODULATION T ECHNIQUES
The process of encoding information by varying one ormore properties of the carrier signal is called modulation.A variety of modulation techniques have been proposed formolecular communications networks by changing the con-centration of molecules, molecules type or the time of themolecules release [5].The first proposed method is the Concentration Shift Keying(CSK) where a bit is decoded at the receiver as ”1” if themolecules concentration reaches a predefined threshold at asymbol interval, otherwise it is a ”0” [37]. Molecular ShiftKeying (MoSK) is another modulation method proposed by thesame authors that uses n molecule types to transmit n bits ofinformation [37]. In [38], the authors proposed Pulse PositionModulation (PPM) that encodes the information by changingthe time of molecules release. They divided the symbol intervalinto two equal halves, and the receiver decodes the bit as”1” if the pulse is in the first half, and as ”0” if the pulseis in the second half. However, the modulation techniquesproposed for molecular communication cannot be used forwired nano-communication networks, which they use electronsas carriers of information instead of molecules. Therefore inthis paper, we propose a new modulation technique for wirednano-communication networks that encodes the informationby changing the intensity of the bioluminescent light, the timebetween two consecutive light emissions or a combinationbetween the intensity and the time changes. We call thesemethods Bioluminescence Intensity Shift Keying (BISK), Bio-luminescence Time Shif Keying (BTSK) and BioluminescenceAsynchronous Shift Keying (BASK) consecutively. A. Bioluminescence Intensity Shift Keying (BISK)
The total emitted light capacity of the designed receiveris proportional to the released Ca concentration and theamount of recycled Aequorin for each symbol interval, whichis proportional to the pulsed electrical intensity transmitted through the nanowire. Therefore, we can modulate the trans-mitted information by varying the intensity of the biolumines-cent light emitted at the receiver. The change in the intensity isdetected at the gateway, where an appropriate threshold is usedto correctly extract the information sent through the nanowire.A bit is decoded as ”1” if the blue light intensity Q e in eq.16 exceeds a threshold ξ , and as ”0” otherwise.The proposed BISK modulation technique has the advan-tage of reducing the Inter-Symbol Interference (ISI) which iscaused by the remaining molecules from a previous symbol.Even if some Ca ions remain in the medium causing theemission of some photons, the intensity of light is still low,and we just need a photo-detector with high resolution at thegateway to eliminate the influence of ISI on the receiver. B. Bioluminescence Time Shift Keying (BTSK)
As in PPM modulation technique proposed in [38], we candivide our symbol interval into two equal halves. The bit isdecoded as ”1” if the bioluminescent light is detected in thefirst half, and as ”0” if it is detected in the second half. Thismethod can be a little bit slower than BISK, because detectingsmall changes in the light intensity is much faster than waitinga hole symbol to decode one bit. However, BTSK can be moreefficient in terms of bit error rate, especially when the majorityof bits have the same value ”1” or ”0”.
C. Bioluminescence Asynchronous Shift Keying (BASK)
Another technique that we propose is a hybrid modulationscheme based on a combination between BISK and BTSK.This modulation technique is asynchronous, and it can achievea higher information rate than the intensity-based and time-based approaches separately. By using k − threshold levels,BASK can use k bits with k different values to representone symbol, which increases the number of bits per symboland thus increases the information rate. Then k = 2 whenwe encode one bit in the intensity chance and another bit inthe time change. Therefore, by using the classical modulationnaming convention based on the number of bits per symbol,BASK modulation technique can be called Quadruple BASK(QBASK) when k = 2 .In this paper we focus on Bioluminescence Intensity ShiftKeying (BISK) modulation technique. The other two proposedmodulation techniques will be covered in our future work. D. Bits Decoding
Since the emission of a bioluminescent light depends onphoto-proteins reacting with the Ca ions released inside thereceiver which exhibit Brownian motion, a single photon hasa certain probability P em of being emitted and detected atthe gateway. This probability p = P em ( r, t b ) defined in eq.16 depends on the number of the recycled photo-protein persymbol interval r and the bit duration t b . Because of the hugenumber of Ca ions released by SER in each bit duration, wecan safely assume that each recycled photo-protein can reactwith three Ca ions at the same time. Thus, the reactionexplained in Fig. 7 is considered as a single event instead of three separate events, whether the photo-protein is activatedto emit a photon or not. We also assume that both Ca ionsand photo-proteins are uniformly distributed inside the receiverand that the bioluminescent reaction happens instantaneously.Based on the assumptions discussed above, if c ions arereleased in a symbol duration, the number of photons emittedby reacting with these c ions is a random variable followinga binomial distribution: C ∼ Binomial ( c, P em ( r, t b )) (17)The Binomial distribution can be approximated to a normaldistribution by Central Theorem Limit (CLT) [39]. When cp islarge enough and p is not closer to zero or one, the distributionof the photons emission variable C can be written as: C n ∼ N ( cp, cp (1 − p )) (18)We assume that the noise inside the designed receiver isAdditive Gaussian White Noise (AWGN). Therefore, the noiseis also a random variable that follows a normal distribution: N ∼ N (0 , σ ) (19)and thus, the probability of detecting the bioluminescent lightemitted at the receiver can be found by adding the two normaldistributions. The probability of detecting a previous bit isneglected in this paper by assuming that the photo-detectorused at the gateway has a high resolution.By using the proposed BISK modulation technique, thegateway decodes the detected bioluminescent light by compar-ing the sum of the two normal distributions with a predefinedthreshold value ξ . If the intended bit is ”0”, then the probabilityof an incorrect decoding for the BISK modulation techniqueare calculated as follows: P r = 13 hcλ P ( C n + N (cid:62) ξ ) (20)and if the intended bit is ”1”: P r = 13 hcλ P ( C n + N < ξ ) (21)VI. P ERFORMANCE E VALUATION AND N UMERICAL R ESULTS
To evaluate the performance of the designed receiver,we used Simulink in MATLAB to simulate the proposedequivalent circuit. The parameters we used in this study arementioned in the section above. The output at the capacitancerepresents the released Ca ion concentration by SER foreach bit interval. As shown in Fig. 9, the current sent throughthe nanowire excites SERs membrane and increases its restingpotential (-75 mV) calculated in eq. 2 by 35 mV, which allowsthe opening of Ca channels. The output of the capacitanceincreases rapidly to reach more than 7 µM of Ca ions, thenwhen the excitation stops, SER starts absorbing the releasedCa ions and the output decreases slowly until it vanishesfrom the medium. Fig. 9. Bioluminescence Intensity Shift Keying (BISK) modulation technique.
A. Number of Ca Channels
In this study, we used the minimum possible channelsnumber per µm area in SER membrane, which is n = 4 channels. These 4 channels represent 4 resistances linked inparallel in our proposed equivalent circuit. We simulated 4cases where only one channel is opened, two, three and fourchannels opened together and we plotted the current necessaryfor each case along with the output of the equivalent circuitas a function with time.In Fig. 10, we can see that the number of open channels isproportional to the intensity of the transmitted current throughthe nanowire. We can also notice that only an intensity of 4pA is needed to open the maximum number of channels weused in this study. This is due to the fact that the membrane’sthickness is so small (6 nm) that a tiny excitation can changethe voltage at the channels and open them. The sensitivity ofSERs membrane to the electric current is one of the reasonswhy we use it to detect electrons in our designed receiver fora wired nano-communication.Fig. 11 shows the output results of the proposed equivalentcircuit in the 4 studied cases. We can notice that with 4 openedchannels, SER can release more than 20 µM of Ca ions.We can also notice that in the case where 4 channels are Fig. 10. The current necessary to open the Ca2+channels. Fig. 11. The released Ca2+ ion concentration foreach number of open channels. Fig. 12. Radiant energy emitted by the biolumi-nescent reaction for different inputs. open, the signal decreased rapidly (1.5 µs ) compared to thecase where only one channel is open (3 µs ). With 4 openedchannels, the SER needs half the time that one channel needsto absorb the released Ca ions after the excitation is stopped,which deceases the intersymbol interference between previousand next bits. The released Ca concentration by SER isproportional to the intensity of the current sent through thenanowire, which is used to demodulate the transmitted signal. B. Radiant Energy
We used the simulated output of the equivalent circuit whichrepresents the concentration of the released Ca ions foreach bit interval c ( t ) to calculate the bioluminescent radiantenergy by using eq. 11. We used different voltage intensities atthe input to show the proportionality between the transmittedcurrent through the nanowire and the light intensity emittedby the receiver. Fig. 12 shows that the intensity of the bluelight emitted by the receiver can reach more than 10 MJ,which is enough to be detected by the gateway. It also showsthat the intensity of light decreases rapidly when the mediumis emptied of Ca ions, because without the presence ofCa ions, the bioluminescent light cannot be emitted, whichdecreases the intersymbol interference.Fig. 13 summarizes the performance of the designed re-ceiver by following the electrons detection and the light emis-sion processes. The figure shows a random binary messagethat is detected by SER, where the bits ”1” open the channelsand release Ca ions, which they combine with Aequorin and emit a blue light. Bit ”0” keep the channels closed, SERabsorbs the Ca ions from the medium and the light emissionstops. The results of this study show that light emission allowsthe designed receiver to play the role of a relay between thenanomachines and the gateway. This ability can be used asa link between the nanoscale and macroscale in the in-bodytechnologies for medical applications. C. Bit Error Rate
In order to evaluate the performance of the designed re-ceiver, we calculate the Bit Error Rate (BER), which is thebit errors number divided by the total transmitted bits duringthe studied time interval. Fig. 14 shows the results of thecalculated BER as function of the detection threshold for the 4studied cases. We notice that the case with 4 open channels has
Fig. 13. The response of the designed receiver to a random binary message. the best results with BER of − for a threshold 16, because 4channels absorb the released Ca ions faster, which deceasesthe intersymbol interference as we explained above. The feweropened channels at the membrane of SER the more bit errorswill be received, where 3 opened channels can reach BER of − for a threshold 17. The worst case is where only onechannel is opened with 0.8, which is explained by the factthat one single channel will not be able to release and absorbCa ions rabidly, which increases intersymbol interferences, Fig. 14. Bit error rate of the designed receiver as a function of the detectionthreshold. and thus, increases the BER.VII. C
ONCLUSION
Using the ability of certain polymers to self-assemble inorder to build conductive nanowires between nanomachinesis a new method proposed for establishing wired nano-communication systems. Despite the very high achievablethroughput of wired nano-communication systems due to theuse of electrons as information carriers, the detection ofelectrons on the nanometric scale is very challenging. Thispaper proposes a bio-inspired receiver design that uses SmoothEndoplasmic Reticulum (SER) to detect the transmitted elec-trons by measuring the released Ca ions concentration withthe photo-protein Aequorin that emits a blue light in thepresence of Ca ions. To better design the receiver, wesimulate the construction of the nanowire, present its electricalcharacteristics and calculate its maximum throughput. Wemodeled the dynamic of SER with an equivalent circuit, andwe derived the analytical expression of their components.We calculated the radiant energy emitted in each symbolinterval, and we simulated the detection and light emissionprocesses of a random binary message. We also proposedmodulation techniques that guaranty an effective decoding ofthe information and we calculated the BER of the designedreceiver to evaluate its performance. The results of this studyshowed that the designed receiver is efficient for wired nano-communication networks and that it can also play the role ofa relay for the nearest gateway.R EFERENCES[1] I. F. Akyildiz and J. M. Jornet, “The Internet of nano-things,”
IEEEWireless Communications , vol. 17, no. 6, pp. 58–63, Dec. 2010.[2] G. Piro, K. Yang, G. Boggia, N. Chopra, L. A. Grieco, and A. Alo-mainy, “Terahertz Communications in Human Tissues at the Nanoscalefor Healthcare Applications,”
IEEE Transactions on Nanotechnology ,vol. 14, no. 3, pp. 404–406, May 2015.[3] S. DOro, L. Galluccio, G. Morabito, and S. Palazzo, “A timing channel-based MAC protocol for energy-efficient nanonetworks,”
Nano Commu-nication Networks , vol. 2, no. 6, pp. 39–50, 2015.[4] X.-W. Yao and J. M. Jornet, “TAB-MAC: Assisted beamforming MACprotocol for Terahertz communication networks,”
Nano CommunicationNetworks , vol. 9, pp. 36–42, Sep. 2016. [5] N. Farsad, H. B. Yilmaz, A. Eckford, C. B. Chae, and W. Guo, “AComprehensive Survey of Recent Advancements in Molecular Commu-nication,”
IEEE Communications Surveys Tutorials , vol. 18, no. 3, pp.1887–1919, 2016.[6] M. Pierobon and I. F. Akyildiz, “A physical end-to-end model formolecular communication in nanonetworks,”
IEEE Journal on SelectedAreas in Communications , vol. 28, no. 4, pp. 602–611, May 2010.[7] N. R. Kim, A. W. Eckford, and C. B. Chae, “Symbol Interval Optimiza-tion for Molecular Communication With Drift,”
IEEE Transactions onNanoBioscience , vol. 13, no. 3, pp. 223–229, Sep. 2014.[8] O. A. Dambri and S. Cherkaoui, “Design Optimization of a MIMOReceiver for Diffusion-based Molecular Communication,” in to appearin Proc. IEEE WCNC 2019, April 14-18 , 2019.[9] B. H. Koo, C. Lee, H. B. Yilmaz, N. Farsad, A. Eckford, and C. B.Chae, “Molecular MIMO: From Theory to Prototype,”
IEEE Journal onSelected Areas in Communications , vol. 34, no. 3, pp. 600–614, Mar.2016.[10] A. Noel, K. C. Cheung, and R. Schober, “Improving Receiver Perfor-mance of Diffusive Molecular Communication With Enzymes,”
IEEETransactions on NanoBioscience , vol. 13, no. 1, pp. 31–43, Mar. 2014.[11] O. A. Dambri and S. Cherkaoui, “Performance enhancement ofdiffusion-based molecular communication,”
IEEE Transactions onNanoBioscience , vol. 19, no. 1, 2019.[12] B. Tepekule, A. E. Pusane, H. B. Yilmaz, C. B. Chae, and T. Tugcu, “ISIMitigation Techniques in Molecular Communication,”
IEEE Transac-tions on Molecular, Biological and Multi-Scale Communications , vol. 1,no. 2, pp. 202–216, Jun. 2015.[13] G. Chang, L. Lin, and H. Yan, “Adaptive Detection and ISI Mitigationfor Mobile Molecular Communication,”
IEEE transactions on nanobio-science , vol. 17, no. 1, pp. 21–35, 2018.[14] O. A. Dambri, S. Cherkaoui, and B. Chakraborty, “Design and Eval-uation of Self-Assembled Actin-Based Nano-Communication,” in , Jun. 2019, pp. 208–213.[15] O. A. Dambri and S. Cherkaoui, “Enhancing Signal Strength and ISI-Avoidance of Diffusion-based Molecular Communication,” in , Jun. 2018, pp. 1–6.[16] A. Noel, Y. Deng, D. Makrakis, and A. Hafid, “Active versus Passive:Receiver Model Transforms for Diffusive Molecular Communication,”in , Dec.2016, pp. 1–6.[17] X. Bao, J. Lin, and W. Zhang, “Channel modeling of molecularcommunication via diffusion with multiple absorbing receivers,”
IEEEWireless Communications Letters , vol. 8, no. 3, pp. 809–812, 2019.[18] D. Kilinc and O. B. Akan, “Receiver design for molecular communi-cation,”
IEEE Journal on Selected Areas in Communications , vol. 31,no. 12, pp. 705–714, 2013.[19] X. Qian, M. Di Renzo, and A. Eckford, “Molecular communications:Model-based and data-driven receiver design and optimization,”
IEEEAccess , vol. 7, pp. 53 555–53 565, 2019.[20] O. A. Dambri and S. Cherkaoui, “A physical channel model for wirednano-communication networks,” arXiv:2005.12328 [cs.ET] , 2020.[21] G. L. Koch, “The endoplasmic reticulum and calcium storage,”
BioEs-says: News and Reviews in Molecular, Cellular and DevelopmentalBiology , vol. 12, no. 11, pp. 527–531, Nov. 1990.[22] O. Shimomura, “A Short Story of Aequorin,”
The Biological Bulletin ,vol. 189, no. 1, pp. 1–5, Aug. 1995.[23] H. Kaur, S. Kumar, I. Kaur, K. Singh, and L. M. Bharadwaj, “Low-intensity magnetic fields assisted alignment of actin filaments,”
Interna-tional Journal of Biological Macromolecules , vol. 47, no. 3, pp. 371–374, Oct. 2010.[24] C. Hunley, D. Uribe, and M. Marucho, “A multi-scale approach todescribe electrical impulses propagating along actin filaments in bothintracellular and in vitro conditions,”
RSC Advances , vol. 8, no. 22, pp.12 017–12 028, 2018.[25] J. A. Tuszyski, S. Portet, J. M. Dixon, C. Luxford, and H. F. Cantiello,“Ionic Wave Propagation along Actin Filaments,”
Biophysical Journal ,vol. 86, no. 4, pp. 1890–1903, Apr. 2004.[26] O. A. Dambri and S. Cherkaoui, “Toward a wired ad hoc nanonetwork,” to appear in Proc. IEEE ICC 2020, Jun 7-12 , Jun. 2020.[27] F. Patolsky, Y. Weizmann, and I. Willner, “Actin-based metallicnanowires as bio-nanotransporters,”
Nature Materials , vol. 3, no. 10,pp. 692–695, Oct. 2004.[28] A. L. Plant, M. Gueguetchkeri, and W. Yap, “Supported phospho-lipid/alkanethiol biomimetic membranes: insulating properties,”
Bio-physical Journal , vol. 67, no. 3, pp. 1126–1133, Sep. 1994. [29] W. Hoiles, V. Krishnamurthy, and B. Cornell, Dynamics of EngineeredArtificial Membranes and Biosensors , 1st ed. Cambridge UniversityPress, 2018.[30] M. Brini, T. Cal, D. Ottolini, and E. Carafoli, “Intracellular calciumhomeostasis and signaling,”
Metal ions in life sciences , vol. 12, p.119168, 2013.[31] A. M. K. MD,
Physiology of the Heart , fifth edition ed. LippincottWilliams and Wilkins, 2010.[32] P. Du, S. Li, G. O’Grady, L. K. Cheng, A. J. Pullan, and J. D. Z. Chen,“Effects of electrical stimulation on isolated rodent gastric smooth mus-cle cells evaluated via a joint computational simulation and experimentalapproach,”
American Journal of Physiology-Gastrointestinal and LiverPhysiology , vol. 297, no. 4, pp. G672–G680, 2009.[33] I. Weeks, L. J. Kricka, and D. Wild, “Chapter 3.2 - signal generationand detection systems (excluding homogeneous assays),” in
The Im-munoassay Handbook (Fourth Edition) , fourth edition ed., D. Wild, Ed.Elsevier, 2013, pp. 267 – 285.[34] O. Shimomura and F. H. Johnson, “Properties of the bioluminescentprotein aequorin,” vol. 8, no. 10, pp. 3991–3997, publisher: AmericanChemical Society.[35] C. Orchard and F. Brette, “t-tubules and sarcoplasmic reticulum functionin cardiac ventricular myocytes,” vol. 77, no. 2.[36] O. A. Krishtal, V. I. Pidoplichko, and Y. A. Shakhovalov, “Conductanceof the calcium channel in the membrane of snail neurones.”
The Journalof Physiology , vol. 310, no. 1, pp. 423–434, 1981.[37] M. . Kuran, H. B. Yilmaz, T. Tugcu, and I. F. Akyildiz, “Interferenceeffects on modulation techniques in diffusion based nanonetworks,”
Nano Communication Networks , vol. 3, no. 1, pp. 65–73, Mar. 2012.[38] N. Garralda, I. Llatser, A. Cabellos-Aparicio, E. Alarcn, and M. Pier-obon, “Diffusion-based physical channel identification in molecularnanonetworks,”
Nano Communication Networks , vol. 2, no. 4, pp. 196–204, 2011.[39] D. J. Wilkinson,