PHLAME: A Physical Layer Aware MAC Protocol for Electromagnetic Nanonetworks

Transcription

1 t IEEE International Workhop on Molecular and Nano Scale Communication (MoNaCom) PHLAME: A Phyical Layer Aware MAC Protocol for Electromagnetic Nanonetwork Joan Capdevila Pujol, Joep Miquel Jornet and Joep Sole Pareta NaNoNetworking Center in Catalunya (N3Cat) Univeritat Politècnica de Catalunya Jordi Girona -3, 834, Barcelona, Spain pareta@ac.upc.edu Broadband Wirele Networking Laboratory (BWN Lab) School of Electrical and Computer Engineering Georgia Intitute of Technology, Atlanta, Georgia 333, USA {joancp, jmjornet}@ece.gatech.edu Abtract Nanotechnology i enabling the development of integrated device jut a few hundred nanometer in ize. Communication among thee nano-device will boot the application of nanotechnology in the biomedical, environmental and military field. Within the communication alternative at the nanocale, the tate of the art in nanomaterial reearch point to the Terahertz band (.- THz) a the frequency range of operation of graphene-baed electromagnetic (EM) nano-tranceiver. Thi frequency band upport very large tranmiion bit-rate and enable imple communication mechanim uited to the limited capabilitie of nano-device. Due to an expectedly very large number of nano-device haring the ame channel, it i neceary to develop new Medium Acce Control (MAC) protocol which will be able to capture the peculiaritie of nanonetwork in the Terahertz band. In thi paper, PHLAME, a phyical layer aware MAC protocol for electromagnetic nanonetwork, i introduced. Thi protocol i built on top of a novel communication cheme baed on the exchange of femtoecond-long pule pread in time, and exploit the benefit of novel low-weight channel coding cheme. In the PHLAME protocol, the tranmitting and receiving nano-device jointly elect the communication parameter that minimize the interference in the nanonetwork and maximize the probability of uccefully decoding the received information. The performance of the protocol i analyzed in term of energy conumption, delay and achievable throughput, by taking alo into account the energy limitation of nano-device. The reult how that, depite it implicity, the PHLAME protocol i able to upport denely populated nanonetwork by exploiting the peculiaritie of the Terahertz band. Index Term Nanonetwork, Terahertz Band, Medium Acce Control, Pule-baed Communication, Graphene I. INTRODUCTION Nanotechnology i providing a new et of tool to the engineering community to deign and manufacture integrated nano-device in a cale ranging from one to a few hundred nanometer. The tak that thee device can individually accomplih are very limited both in term of complexity and range of operation. By mean of communication, nanodevice will be able to achieve more complex tak in a ditributed manner and to cover larger area [], []. The Thi work wa completed during hi tay in the Broadband Wirele Networking Lab. reulting nanonetwork, i.e., network of nano-device, will expand the range of application of nanotechnology in the biomedical, environmental and military field, amongt other. For the time being, the communication alternative for nanodevice are very limited. Focuing on the electromagnetic (EM) paradigm, the utilization of novel nanomaterial, uch a graphene, i enabling the development of miniaturized EM tranceiver uited to the target ize and energy capabilitie of nano-device [3], [4]. Amongt other, ongoing reearch on the characterization of the EM propertie of graphene [5], [6], [7] point to the Terahertz band (.-. THz) a the expected frequency range of operation of future EM nanotranceiver. In particular, in [5] we determined that a µm long graphene-baed nano-antenna can only efficiently radiate in the Terahertz range. Thi matche the prediction for the frequency of operation of graphene-baed RF tranitor [8]. The Terahertz band (.-. THz), on it turn, i one of the leat explored frequency range [9]. In [], we developed a propagation model for Terahertz communication and howed how the Terahertz band can theoretically upport very large bit-rate, up to everal hundred of terabit per econd for ditance below one meter. However, it i not likely that nanodevice will require thee very large tranmiion bit-rate in many application. Alternatively, having a very large bandwidth alo enable new imple communication mechanim uited to the expectedly limited capabilitie of nano-device. We alo introduced a new communication cheme for nanodevice baed on the exchange of very hort pule pread in time in []. Indeed, due to the ize and energy contraint of nano-device, it i currently not feaible to generate a high-power carrier ignal in the nanocale at Terahertz frequencie []. A a reult, claical communication paradigm baed on the tranmiion of continuou ignal cannot be ued. On the other hand, very hort pule can be generated and efficiently radiated in the nanocale [6]. In particular, femtoecond-long pule, which have their main frequency component in the Terahertz band, are already being ued in everal application uch a nanocale imaging []. In light of the very large number of nano-device and the //$6. IEEE 43

2 random nature of nanonetwork [3], there i a need for new Medium Acce Control (MAC) protocol. It i not practical to ue the claical olution becaue they do not capture the peculiaritie of nanonetwork. Firt, the main limitation at the nanocale i not the available bandwidth, but the energy of nano-device, which can only be provided by mean of energy harveting ytem [4]. Second, claical MAC protocol are not directly applicable in pule-baed communication ytem. Only ome of the olution propoed for Impule Radio Ultra Wide Band (IR-UWB) network [5] could be conidered, but their complexity limit their uefulne. In thi paper, we preent PHLAME, a PHyical Layer Aware MAC protocol for Electromagnetic nanonetwork. The PHLAME protocol i baed on the joint election by the tranmitter and the receiver of the communication parameter and the channel coding cheme that minimize the interference in the nanonetwork and maximize the probability of uccefully decoding the received information. To the bet of our knowledge, thi i the firt MAC protocol for EM nanonetwork that capture the peculiaritie of the Terahertz band a well a the expected capabilitie of future graphenebaed nano-device. Our main contribution are: We preent Rate Diviion Time Spread On-Off Keying, RD TS-OOK, a revied verion of the communication cheme baed on the exchange of femtoecond-long pule that we firt introduced in [], in order to upport different ymbol and coding rate. We propoe a phyical-aware MAC protocol for EM nanonetwork, PHLAME, a new channel haring protocol that adapt the RD TS-OOK coding parameter according to the tranmitter and receiver perceived channel quality and available reource. We analyze the performance of the propoed protocol by mean of energy conumption, delay and achievable throughput, by uing accurate model of the Terahertz channel (path-lo and molecular aborption noie) and the interference. The ret of thi paper i organized a follow. In Sec. II, we decribe the new pule-baed communication cheme which i conidered in our analyi. In Sec. III, we preent our new MAC protocol for EM nanonetwork and highlight the noveltie of thi olution. In Sec. IV, we invetigate the performance of the preented protocol in term of energy conumption, delay and throughput. Finally, we conclude the paper in Sec. V. II. RATE DIVISION TIME SPREAD ON-OFF KEYING The Rate Diviion Time Spread On-Off Keying communication cheme (RD TS-OOK) i a new modulation and channel haring mechanim for nano-device baed on the aynchronou exchange of femtoecond-long pule, which are tranmitted following an on-off keying modulation pread in time. A implified verion of thi mechanim wa firt introduced in []. The functioning of thi communication cheme i a follow. Auming that a nano-device need to tranmit a binary tream (e.g., the reading from a nanoenor), Tx Tx Rx τ τ t prop T p T T p t prop T 5 5 Time [p] Fig.. Rate Diviion TS-OOK. A logical i tranmitted by uing a femtoecondlong pule and a logical i tranmitted a ilence, i.e., the nano-device remain ilent when a logical zero i tranmitted. An On-Off Keying (OOK) modulation, baed on the preence or abence of radiation, i choen intead of a binary Pule Amplitude Modulation (PAM), baed on the change of the ignal polarity, becaue of the peculiar behavior of molecular aborption noie in the Terahertz band. Thi type of noie i trongly preent when molecule are excited []. The time between ymbol T i much longer than the pule duration T p,anditifixedforthedurationofa packet. Due to technology limitation and imilarly to Impule Radio Ultra-Wide-Band (IR-UWB) ytem [5], the ymbol, i.e., the pule or the ilence, are not tranmitted in a burt, but pread in time. By determining the time between ymbol, i.e., the ymbol rate, after the detection of the firt tranmitted pule a uer doe not need to continuouly ene the channel. The time between pule, i.e., the ymbol rate, β, i different for different uer and different type of packet. Even if unlikely, very hort pule can collide. If the nano-device are tranmitting at the ame rate, a colliion in one ymbol entail a colliion in every ymbol until the end of the packet, which are uually referred to a catatrophic colliion. In other pule-baed cheme uch a in IR-UWB, orthogonal time hopping equence are ued to avoid thi condition [5]. Due to the complexity of generating thee equence in nano-device, we advocate for the variation of the ymbol rate [6]. Several nano-device can concurrently occupy the channel when uing RD TS-OOK mainly due to the fact that the time between ymbol T i much longer than the ymbol duration T p. The tranmiion of very hort pule (le than femtoecond []) minimize the chance of having colliion, and provide almot orthogonal communication channel. Note that under thi communication cheme, a colliion between packet only occur when two or more ymbol exactly overlap in time. Moreover, by allowing different uer to tranmit at different rate, a colliion in a given ymbol doe not lead to multiple conecutive colliion in the ame packet. Fig. how the RD TS-OOK ignal tranmitted by two uer with different initial tranmiion time τ and τ.the upper plot correpond to the equence, which i tranmitted by the firt uer. A logical i repreented by ahortpuleandalogical irepreentedbyilence.the time between ymbol, T i much larger than the ymbol 43

3 duration T p.thitranmittedignalipropagatedthroughthe channel and corrupted with molecular aborption noie by the time it reache the receiver. Similarly, the econd plot how the equence tranmitted by the econd uer,, with a different ymbol rate T.Inthiexample,theecondueri farther from the receiver than the firt uer. A a reult, the ignal at the receiver uffer from higher attenuation, longer delay, and more noie. The ignal at the receiver ide, R t, i hown in the third plot. In thi pecific cae, the delay introduced by the channel to each ignal, t prop and t prop, i uch that the firt ymbol of the econd uer overlap with the econd ymbol of the firt uer. A a reult of uing different rate, conecutive ymbol in both uer do not overlap. III. A PHYSICAL LAYER AWARE MAC PROTOCOL FOR NANONETWORKS We introduce the PHLAME protocol a the firt MAC protocol which i tailored to the peculiaritie of the Terahertz band and which take into account realitic limitation of future nanoelectronic device. The protocol i built on top of RD TS-OOK, and it i plit in two tage, namely, the handhaking proce and the data tranmiion proce. A. Handhaking Proce The aim of the handhaking proce i twofold. Firt, it allow a receiver to coordinate multiple imultaneou tranmiion. Second, it facilitate the joint election of both the tranmiion ymbol rate and the channel coding cheme that make the data tranmiion more reliable. The handhaking proce i divided in two ubtage, the handhaking requet and the handhaking acknowledgment. The handhaking requet i triggered by any nano-device that ha information to be tranmitted and which ha enough energy to complete the proce. A tranmitter generate a Tranmiion Requet (TR) packet, which contain the Synchronization Trailer, the Tranmitter ID, the Receiver ID, the Packet ID, the tranmitting Data Symbol Rate (DSR) and the Error Detecting Code (EDC). The DSR field pecifie the ymbol rate β that will be ued to tranmit the data packet. The trength of RD TS-OOK againt colliion increae when different uer tranmit at different rate. In the PHLAME protocol, every tranmitting node randomly elect a ymbol rate from a et of coprime rate, which minimize the probability of having catatrophic colliion [6]. The EDC field i ued to detect tranmiion error a a conventional checkum field. The TR packet i tranmitted uing a Common Coding Scheme (CCS), which pecifie a predefined ymbol rate and channel coding mechanim. By uing the ame ymbol rate, catatrophic colliion might occur. However, the TR packet are very hort and the EDC field hould uffice to detect imple error in the majority of cae. Finally, the tranmitter wait for a timeout before trying to retranmit the TR packet when no anwer i received. The handhaking acknowledgment i triggered by the receiver of the TR packet, which ue the CCS to decode the received bittream when litening to the channel. If a TR packet i uccefully decoded, the receiver will check whether it can handle an additional incoming bittream. In our cenario, we conider that due to the energy limitation of nano-device, after the tranmiion or active reception of a packet, a device need to wait for a certain recovery time in order to retore it energy by mean of energy harveting ytem [4]. Thi time i much longer than the packet tranmiion delay and poe a major limitation to the network. If the handhake i accepted, a Tranmiion Confirmation (TC) packet i ent to the tranmitter uing the CCS. The TC packet contain the Synchronization Trailer, the Tranmitter ID, the Receiver ID, the Packet ID, the tranmitting Data Coding Scheme (DCS) and the Error Detecting Code. The DCS i elected by the receiver in order to guarantee a target Packet Error Rate (PER), which depend on the perceived channel quality and can be etimated from the pule intenity or the perceived noie. In particular, the DCS determine two parameter value. Firt, it pecifie the channel code weight, i.e., the average number of logical in the encoded data. By reducing the code weight, interference can be mitigated without affecting the achievable information rate, a we howed in []. Second, the DCS pecifie the order of the repetition code that will be ued to protect the information. Since RD TS-OOK reduce poible tranmiion error by avoiding catatrophic ymbol colliion, a imple repetition code i enough to uccefully decode the information in the majority of cae. B. Data Tranmiion Proce At thi point, the data i tranmitted at the ymbol rate pecified by the tranmitter in the DSR field, and encoded with the weight and repetition code pecified by the receiver in the DCS field. The DP contain a Synchronization Trailer, the Tranmitter ID, the Receiver ID, and the ueful Data. The Error Detecting Code ha been removed from the packet ince by uing different ymbol rate, catatrophic colliion are highly unlikely, and randomly poitioned error can be fixed by mean of the choen channel coding cheme. If the DP i not detected at the receiver before a time-out, Tout DP,thereceiveraume that the handhaking proce failed. IV. PERFORMANCE ANALYSIS In thi ection, we analyze the performance of the PHLAME protocol in term of energy conumption, packet latency and normalized throughput. A. Sytem Model The following aumption are conidered in our analyi: The path-lo and noie in the Terahertz band are computed by uing the model introduced in []. A tandard medium with % of water vapor i conidered. The interference i modeled a in [], by auming a Poion field of interferer. The denity of active node i a parameter value in our analyi. The tranmitter encode logical by uing the firt time-derivative of femtoecond long Gauian pule. The energy of a pule i limited to pj. Anon-coherentreceiverarchitectureiconidered,with an integration time T i equal to ten time the ymbol duration T p [7]. 433

4 The recovery time for a nano-device after tranmiion or active reception of a DP i three order of magnitude longer than the data packet duration [4]. The receiver can imultaneouly track a fixed number of incoming packet, K. We model thi a a finite length queueing ytem with K erver and without waiting lane (a packet that cannot be erved i dicarded) [8]. The RD TS-OOK ymbol rate are randomly choen by each node from a pool of pairwie coprime rate code in the order of (e.g., 9, 3, 9). The TR and TC packet in the PHLAME protocol are 6 Byte. DP are 5 Kbyte. The packet length i arbitrarily choen, but it eem appropriate to ue relatively large DP becaue RD TS-OOK doe not cannibalize the channel and tranmiion error are expectedly pare. The target Packet Error Rate i equal to 3. The poible bit coding cheme are limited to a non repetition code with weight equal to.5 (the number of logical and i the ame), a 3-repetition code with weight equal to.4 (only 4% of the bit are logical ), a 5-repetition code with weight equal to.3, a 7-repetition code with weight equal to. and a 9-repetition code with weight equal to.. We undertand by a n-repetition code a coding cheme that replicate n time each ymbol, either pule or ilence. B. Energy Conumption The energy conumption i contributed by the conumption at the tranmitter and at the receiver. Currently, the energyconumption of graphene-baed nano-electronic i till unknown. Becaue of thi, we focu on the energy that would be pent only in the communication part. Thee reult hould be caled by the overall efficiency of a graphene-baed nano-tranceiver. ) Tranmitter Energy Conumption: Thi i mainly governed by the number of handhaking attempt and the length and code weight ued for the tranmiion of the DP. Three poible cae can happen when tarting a new packet tranmiion. Firt, the handhaking can fail becaue the TR packet collide with other packet, or becaue the receiver either cannot allocate one more tranmiion or it i in it energy recovery tage. Second, the handhaking can be aborted becaue the TC packet collide at the tranmitter. The third cae correpond to the ituation in which the handhaking ucceed, and the node go into the Data Tranmiion phae. To etimate the energy conumption at the tranmitter, we conider the energy involved in the tranmiion, E, reception, E, and time-out, E t o, for each one of the aforementioned cae. Thee partial energie are given by: E E E 3 E TR E TR E TR Et H o E TC E TC E DP. () Each type of packet ued by the PHLAME protocol ha a different number of bit and i encoded uing different channel coding cheme. Moreover, the data packet tructure depend on the elected DSR and DCS. When more robut code are needed, the repetition code order i increaed and it weight i reduced. Thi make packet longer but not necearily much more energy conuming, becaue only the tranmiion of pule conume energy, and thi decreae with the code weight. At the ame time, tranmitting with lower weight code can alo reduce the overall interference and ultimately the number of retranmiion []. Then, each cae for the energy conumption decribed above occur with a certain probability, which can be calculated a: p a ptr p p 3 a a p TR p TC p TR p TC where p a refer to the probability of acceptance at the receiver, and p refer to the probability of ucceful reception. The p a i computed by taking into account the maximum number of imultaneou incoming packet that the receiver can handle K and it energy tatu. The p i computed from the probability of ymbol error for the Terahertz channel with the type of pule that are conidered, and by taking into account the error correcting capabilitie of the channel code in ue. Then, the conumed energy in the tranmitter depend on the number of retranmiion required to complete the handhaking. Since the probability of ucceful handhaking i exactly p 3,theenergyconumedatthetranmitteri: E tranmitter p E p E p 3 E 3. (3) p 3 By combining () and () into (3), we reach the following cloed-form expreion: E tranmitter a ptr p TC a p TR p TC a ptr E D P. E H t o ) Receiver Energy Conumption: The energy at the receiver i governed by the number of handhaking attempt a well a the DP tranmiion. The handhaking fail when the receiving node i unable to decode the TR packet, when it cannot handle another tranmiion or when the TC packet collide. Similarly a before, by expreing the energie and the probabilitie for each cae, the energy conumption at the receiving node can be written a: E receiver a p TR p TC a ptr p TC a p TR E D t o P () (4) E D P. (5) Finally, the total energy conumption per ueful bit of information i obtained by adding (4) and (5) and dividing it by the length of the DP. In Fig. (left), the total energy conumption per bit a a function of the node denity i hown for different maximum number of imultaneouly handled packet at the receiver, k. Whenthenodedenityiincreaed,theinterferencein the network increae, which ha a twofold impact on the 434

5 Enery per bit [pj] 4 3 k= k= k=3 k= Node denity [Node/mm ] Average packet delay [m] k= k= k=3 k= Node denity [Node/mm ] Normalized Throughput Node denity [Node/mm ] Fig.. Energy per bit conumption, average packet delay and normalized throughput a function of the node denity for different maximum number of imultaneou packet that can be handled by the receiver. k= k= k=3 k=4 Enery per bit [µj] w/ Handhaking w/o Handhaking Node denity [Node/mm ] Average packet delay [] 5 5 w/ Handhaking w/o Handhaking Node denity [Node/mm ] Normalized Throughput w/ Handhaking w/o Handhaking Node denity [Node/mm ] Fig. 3. Comparion between PHLAME and imilar protocol without handhaking tage in term of the energy per bit conumption, average packet delay and normalized throughput a function of the node denity for different maximum number of imultaneou packet that can behandledbythereceiver. energy conumption. Firt, a higher interference turn into an increaed number of handhaking attempt. Second, once the handhake ha been completed, the DP i tranmitted uing higher order repetition code which are neceary to guarantee the target PER. The tep in the energy curve correpond to the tranition in the coding cheme from non repetition code to 3-repetition, 5-repetition, and o on. At the ame time, note that by allowing the receiver to handle more than one packet imultaneouly, the energy decreae. In Fig. 3 (left), we how the energy conumption per ueful bit of information in a nanonetwork operating under RD TS-OOK, but in which rather than uing the PHLAME protocol, the DP are directly tranmitted without any type of handhaking. There are almot three order of magnitude of difference between the PHLAME protocol and the protocol without handhake. Thi reult depend on the packet length and the offered load parameter. For a very dene network, a the one we are conidering, a handhake avoid having to retranmit the entire DP everal time. We acknowledge that amorecompleteanalyiontheimpactofthepacketizein the ytem ha to be conducted. Finally, we would like to emphaize the energy reduction achieved by uing low-weight coding cheme. In Fig. 4, the energy conumption per bit of the PHLAME protocol i compared to that of the cae in which only the repetition code order i variable and the code weight remain at.5. The reult how that epecially for very dene network, lowering the code weight can reduce the overall energy conumption by more than half. Thi i due to the fact that the interference i mitigated when uing lower weight code, and thi minimize both the number of handhake attempt and the probability of ymbol error and energy conumed in the DP. C. Packet Latency To tudy the packet latency we hould take into account that the different type of packet in the PHLAME protocol have different length and are encoded uing different parameter. In particular, we conider that packet have the following average duration: T TR B TR β min T i T TC B TC β min T i (6) T DP B DP β max β min N r T i where T TR, T TC and T DP tand for the packet duration of TR, TC and DP packet, repectively, β min and β max are the minimum and maximum ymbol rate that the nano-device can elect, T i refer to the integration time and N r i the required number of ymbol per bit to achieve the target PER. Following a imilar procedure a before, we can write the cloed-form expreion for the average packet delay a: T PCK a ptr T TR Tt H o a ptr p TC a ptr p TC T TR Tt DP o T TR T TC T DP. In Fig. (center), the average packet delay given by (7) i hown a a function of the node denity. The impact of the capabilitie of the receiving node in term of maximum number of packet that a nano-device can handle i illutrated. When the node denity i increaed, the interference i increaed, and conequently the number of handhaking attempt increae. Thi turn into longer packet tranmiion delay. However, the major increae come from the change (7) 435

6 Enery per bit [pj] k= Adaptive Weight k=4 Adaptive Weight k= Weight =.5 k=4 Weight = Node denity [Node/mm ] Fig. 4. Energy per bit conumption a a function of the node denity for different code weight. in the repetition code order that i neceary to achieve the target packet error rate. Similarly a before, by allowing the receiver to handle more than one packet imultaneouly, the overall delay i clearly reduced. Finally, note that a imple handhaking proce can reduce the time delay by almot three order of magnitude, a hown in Fig. 3 (center), where the delay in the PHLAME protocol i compared to that of utilizing RD TS-OOK without handhaking proce. D. Normalized Throughput We define the normalized throughput a the maximum information rate that the MAC layer can upport divided by the maximum data rate that a node can tranmit in a ingle uer cenario. For thi, we divide the uer bit-rate that the PHLAME protocol can provide by the maximum achievable bit-rate impoed by RD TS-OOK. Thi i given by, Tput Rb PHLAME Rb max bp bp L D T PCK βmax β N min r T i (8) where L D tand for the payload length in the data packet, T PCK i the packet latency found in (7), N r refer to the coding rate ued, T i i the obervation time and β max, β min are the maximum and minimum ymbol data rate, repectively. The normalized throughput i hown in Fig. a a function of the node denity. Similarly a before, the change in the coding cheme a the interference increae, create the tep in the throughput curve. A expected, the normalized throughput of the PHLAME protocol i much larger than that of a imilar protocol without the handhaking tage (Fig. 3). The main reaon for thi reult come from the fact that the handhake doe not only inform the receiver about a new incoming tranmiion, but firt, it ak for it permiion baed on it local tatu, and, econd, determine the bet communication parameter and coding cheme. V. CONCLUSIONS In thi paper, we preent a phyical layer aware MAC protocol for electromagnetic nanonetwork, PHLAME. Thi protocol i tailored to a novel communication cheme baed on the exchange of femtoecond-long pule pread in time. Our olution allow the tranmitter and the receiver to jointly elect in an adaptive fahion everal communication parameter uch a the ymbol rate or encoding cheme and channel code weight, by mean of a handhaking proce. We analyze the performance of the propoed protocol in term of energy conumption per ueful bit of information, average packet delay and normalized achievable throughput. The reult how that, depite it implicity, the PHLAME protocol i able to upport denely populated nanonetwork by exploiting the peculiaritie of the Terahertz band, the expected capabilitie of future electronic graphene-baed nano-device, and the benefit of low weight coding cheme. Future work include the invetigation of the impact of the packet ize on the overall network performance, and the validation of thee reult by mean of a network imulation tool. ACKNOWLEDGEMENT Thi work wa upported by the US National Science Foundation (NSF) under Grant No. CNS-9663 and Fundación Caja Madrid. REFERENCES [] I. F. Akyildiz, F. Brunetti, and C. Blazquez, Nanonetwork: A new communication paradigm, Computer Network (Elevier) Journal, vol.5, no., pp. 6 79, Augut 8. [] I. F. Akyildiz and J. M. Jornet, Electromagnetic wirele nanoenor network, Nano Communication Network (Elevier) Journal, vol., no., pp. 3 9, March. [3] P. Avouri, Carbon nanotube electronic and photonic, Phyic Today, vol. 6, no., pp. 34 4, January 9. [4] P. Kim, Toward carbon baed electronic, in IEEE Device Reearch Conference, June8. [5] J. M. Jornet and I. F. Akyildiz, Graphene-baed nano-antenna for electromagnetic nanocommunication in the terahertz band, in Proc. of 4th European Conference on Antenna and Propagation, EUCAP, April, pp. 5. [6] M. Roenau da Cota, O. V. Kibi, and M. E. Portnoi, Carbon nanotube a a bai for terahertz emitter and detector, Microelectronic Journal, vol. 4, no. 4-5, pp , April 9. [7] G. Zhou, M. Yang, X. Xiao, and Y. Li, Electronic tranport in aquantumwireunderexternalterahertzelectromagneticirradiation, Phyical Review B, vol.68,no.5,p.5539,october3. [8] Y. M. Lin, C. Dimitrakopoulo, K. A. Jenkin, D. B. Farmer, H. Y. Chiu, A. Grill, and P. Avouri, -GHz Tranitor from Wafer-Scale Epitaxial Graphene, Science, vol.37,no.5966,p.66,february. [9] IEEE 8.5 Wirele Peronal Area Network - Terahertz Interet Group (IGthz). [Online]. Available: [] J. M. Jornet and I. F. Akyildiz, Channel capacity of electromagnetic nanonetwork in the terahertz band, in Proc. of IEEE International Conference on Communication, ICC, May,pp. 6. [], Low-weight channel coding for interference mitigation in electromagnetic nanonetwork in the terahertz band, in to appear in Proc. of IEEE International Conference on Communication, ICC, June, pp. 6. [] D. Woolard, P. Zhao, C. Rutherglen, Z. Yu, P. Burke, S. Brueck, and A. Stintz, Nanocale imaging technology for thz-frequency tranmiion microcopy, International Journal of High Speed Electronic and Sytem, vol.8,no.,pp.5,8. [3] I. F. Akyildiz and J. M. Jornet, The Internet of Nano-Thing, IEEE Wirele Communication Magazine, vol.7,no.6,pp.58 63,December. [4] Z. L. Wang, Toward elf-powered nanoytem: From nanogenerator to nanopiezotronic, Advanced Functional Material, vol. 8, no., pp , 8. [5] IEEE 8.5.4a: Wirele Medium Acce Control (MAC) and Phyical Layer (PHY) Specification for Low-Rate Wirele Peronal Area Network (WPAN). Amendment : Add Alternate PHY., IEEEStandard for Information Technology, Telecommunication and Information Exchange between Sytem Std. [6] M. Weienhorn and W. Hirt, Uncoordinated rate-diviion multipleacce cheme for puled uwb ignal, IEEE Tranaction on Vehicular Technology,vol.54,no.5,pp ,September5. [7] A. Goldmith, Wirele Communication. New York, NY, USA: Cambridge Univerity Pre, 5. [8] L. Kleinrock, Queueing Sytem. Volume : Theory. Wiley-Intercience,