Indonesian Journal of Electrical Engineering and Computer Science V ol. 7, No . 3, September 2017, pp . 802 808 DOI: 10.11591/ijeecs .v7.i3.pp802-808 802 Optimal Ener gy Har vesting Sc heme f or P o wer Beacon-Assisted Wireless-P o wered Netw orks Dinh-Thuan Do F aculty of Electronics T echnology , Industr ial Univ ersity of Ho Chi Minh City 12 Nguy en V an Bao , Go V ap Dist., Ho Chi Minh City , Vietnam e-mail: dodinhthuan@gmail.com Abstract In this paper , w e consid er one-w a y rela y with energy har v esting system based on po w er beacon (PB), in which the rela y node har v ests tr ansmitted po w er from the PB station to f orw ard signals to destination. W e also analyse the rela y netw or k model with amplify-and-f orw ard (AF) protocol f or inf or mation cooper ation and P o w er Splitting-based Rela ying (PSR) protocol f or po w er tr a nsf er . In par ticular , t he outage probability and optimal energy har v esting (EH) po w er splitting fr action of no v el scheme in are p resented. W e obtain analytical closed-f or m e xpression of optimal energy har v esting (EH) po w er splitting fr action to minimiz e the outage probability of system. Using n umer ical and analytical sim ulations , the perf or mances of diff erent cases are presented and discussed. K e yw or ds: po w er beacon, Energy Har v esting, po w er splitting-based rela ying, Amplify and F orw ard Cop yright c 2017 Institute of Ad v anced Engineering and Science . All rights reser v ed. 1. INTR ODUCTION Recently , wireless access will be a reality in fifth gener ation (5G) wireless systems , with a ser ies of emerging technology , such as massiv e m ultiple-input and m ultiple-output (MIMO), de vice-to-de vice comm unication and small cell architectures , which ha v e launched a huge data v olume in wireless ser vices , such as mobile gaming , mobile TV and mobile Inter net. Fur ther more , with f ast increasing n umber of users o wning smar tphones and tab lets , and hence one of the most cr itical prob lem is that limited oper ation lif e time of mobile de vices due to finite energy ability . T o address this hard issue , r adio-frequency (RF) based energy har v esting model has receiv ed a consider ab le research interest [1]. A promising technology of RF-assisted energy signals enab les wireless energy tr ansf er (WET) to pro vide contin uous and con v enient po w er supply to energy- a w are wire less ter minals . As a result, wireless po w ered comm unication (WPC) system together with wireless de vices (WDs) is designed to tr ansf er data to the inf or mation receiv ers (IRs), e .g., sensor nodes . Such ter minals is po w ered up b y the means of wireless po w ered equipment as a dedicated energy tr ansmitters (ETs). In f act, wireless recharging is one of the most desir ab le ne w f eatures f or mobile de vices to eliminate the need of po w er cords and po w er charge rs . T o realiz e this wireless po w er tr ansf er , a no v el netw or k architecture where stations called po w er beacons (PBs) are installed in tr aditional cellular netw or k f or boosting oper ation efficiency of mobiles and sensors via micro w a v e r adiation kno wn as micro w a v e po w er tr ansf er (MPT) In [2][9], the w or ks ha v e lately adopted some tr ansmission policies f or EH cooper ativ e netw or ks . EH rela ys w ere first imp lemented in cooper ativ e comm unication [2] . In [3][5], assuming a perf ect EH model under which the energy arr iv al time and the amount of har v ested energy are calculated pr ior to tr ansmit such energy , regarding EH rela y systems , there w ere some po w er allocation policies giv en. Ne v er theless , due to the r andom energy arr iv al time and the amount of har v ested energy , the dete r ministic EH model seems to be abstr act. Thus , under gener al energy har v esting profiles , the authors in [6][9] proposed a n umber of tr ansmission policies . P ar ticular ly , there w ere se v er al joint rela y selection and po w er allocation schemes giv en due to th e stationar ity and erodicity of the energy har v esting process [6]. Lik e wise , in [7], [8], since dur ing an y time of data tr ansmission, energy can be sca v enged, se v er al po w er allocation schemes f or cooper ativ e EH netw or ks w ere put f orw ard. The study in [9] based on what ha v e been accomplished in [7] Receiv ed Apr il 22, 2017; Re vised A ugust 12, 2017; Accepted A ugust 27, 2017 Evaluation Warning : The document was created with Spire.PDF for Python.
IJEECS ISSN: 2502-4752 803 and [8] to e xpand the b uff er-aided link adaptiv e EH rela y system, in which the change of EH r ates can be positiv ely assisted. In [10], the authors planned the no v el impression of de plo ying dedicated po w er nodes to enab le wireless po w er tr ansf er in an e xisting cellular netw or k, named po w er beacons (PBs). The densities and tr ansmit po w er of PBs are in v estigated. By using tool of the stochastic ge- ometr y theor y under data links’ outage constr aint. With this PB based energy tr ansf er , w e thus could consider a ne w netw or k , so-called PB-assisted wireless netw or ks . In such netw or ks , each user can har v est wireless energy from the dedicated PB . F or th is ne w model, the optimization prob lem need be solv ed to optimally allocate the resources of po w ered netw or k including po w er and time fr action f or wireless po w er processing and inf or mation tr ansmission. T o the best of our kno wledge , this is still an open question, which motiv ates this paper . In this paper , w e consider a PB rela ying netw or k consisting of one PB and rela y-destination pair . W e assume that the PB are connected to constant po w er supplies . Each AP (access point) station aims to collect the inf or mation from its associated source . It is assumed that each source has no embedded energy supply b ut has the ability to har v est and store the energy from RF signals broadcast b y its AP . The PB is installed to assist APs dur ing the energy har v esting phase . Specifically , w e f or m ulate a closed-f or m e xpression of the optimal resource po w er allocation of energy har v esting protocol to obtain optimal outage perf or mance . The rest of the paper is organiz ed as f ollo ws . Section II descr ibes the signal and channel models of the EH enab led rela ying netw or k. In Section III, the outage probability and throughput are f or m ulated and solv ed in order to maximiz e perf or mance . Numer ical result s and compar isons are presented in section IV . Finally , summar izing remar ks are giv en in section V . 2. SYSTEM MODEL Figure 1. The system model In this paper , w e consider a PB-assisted rela ying netw or k consisting of one m ulti-antenna PB and rela y-destination pairs . In vie w of the state of ar t and trend of RF energy tr ansf er , the considered netw or k setup is v er y lik ely to find its applications in the pr actical scenar io of small cells , such as picocells (r ange from 10 to 100 meters) and f emtocells (WiFi lik e r ange), which has been regarde d as one of the k e y e nab ling technologies of the upcoming 5G cellular netw or ks . It is also w or th mentioning that although introducing the PB ma y result in some e xtr a cost and comple xity to the system, this could be beneficial as a whole based on the f ollo wing consider ations: (i) The PB could be dedicated designed f or po w er tr ansf er only and thus can achie v e a higher energy har v esting efficiency b y e xploiting the benefits of energy beamf or ming enab led b y m ultiple antennas . (ii) The PB design is adapted to applications required lo w po w er such as in small cell netw or ks . Optimal Energy Har v esting Scheme f or P o w er Beacon-Assisted ... (Dinh-Thuan Do) Evaluation Warning : The document was created with Spire.PDF for Python.
804 ISSN: 2502-4752 (iii) It can be f ound optimal energy le v el f or energy har v esting phase and calculated at PB . As a result, model of PB can pro vide wireless charging ser vices in real lif e . The entire comm unication consists of tw o diff erent phases , namely , energy har v esting and inf or mation tr ansmission. Assuming a po w er fr action of tr ansmitted po w er of PB dur ing the first phase is assigned to energy tr ansf er , in which the AP station har v est energy from the PB to f orw ard signal to destination D . Dur ing the energy har v esting processing stage , t he receiv ed signal at the AP can be e xpressed as y r = p P S h T 1 x + n r ; (1) where n r is the additiv e white Gaussian noise (A WGN) with with the v ar iance as N 0 . It is noted that the PB is equipped with m ultiple antennas , energy beamf or ming is de- plo y ed to prog ress the perf or mance of energy tr ansf er , i.e ., x = ws ; (2) where w is the beamf or ming v ector with w = 1 and s is the energy symbol with unit po w er . There- f ore , the optimal beamf or ming v ector is giv en b y w = h H k h k 2 ; (3) in which ( : ) H denotes the Her mitian tr anspose matr ix. F ollo wing the pr inciple of po w er splitting -based rela y prototcol, the first par t in the re- ceiv ed RF signal is allocated f or energy gather ing with po w er fr action 2 [0 ; 1] . Applying no v el energy har v esting pr inciple , the total e xpected energy at the end of the first phase at the AP can be calculated as E h = P S k h k 2 + N 0 T ; (4) where (0 < 1) is the energy con v ersion efficiency of the EH circuit, T is b loc k time of signal processing per iod, P S is tr ansmit po w er of the PB . The har v ested po w er at the AP use to f orw ard inf or mation to the destination node , e .g. mobile users , this procedure is deplo y ed in the second stage . Ne xt, the tr ansmitted po w er at the AP is e xpressed as P r = E h T = P S k h k 2 + N 0 (5) In pr actical applications , P r is g reater than 0, which results in 0 < 1 . At the same time of the first stage , the RF signal perf or m an allocation as p 1 y r f or inf or mation processing. The receiv ed signal in the inf or mation processing stage is giv en b y ~ y r = p 1 p P S hx + n r + n p ; (6) in which n p denotes is the additional baseband Gaussian noise with z ero mean and v ar iance N 0 . In this scenar io , the po w er splitting fr action is s e lected appropr iately f or obtaining the optimal outage perf or mance . In this model, AF scheme is applied in rela y node , after that the tr ansmitted signal at rela y in the second hop is e xpressed as x r = p P r G ( ) ~ y r : (7) In this proposed architecture , the AP is assumed that it equipped b y amplify circuit to broaden har v ested po w er and then using such po w er f or the ne xt processing stage . The rela y amplifies the receiv ed signal and the tr ansmitted signal from the rela y b y the f actor as belo w G ( ) = v u u t 1 (1 ) P S k h k 2 + N 0 + N 0 (8) IJEECS V ol. 7, No . 3, September 2017 : 802 808 Evaluation Warning : The document was created with Spire.PDF for Python.
IJEECS ISSN: 2502-4752 805 In the second phase , AP tr ansmits inf or mation to mobile user via channel link using the energy har v ested in the first phase . Hence , the receiv ed signal As a result, receiv ed signal at mobile user is f or m ulated b y y d = g x r + n d = r (1 ) P S P r (1 ) ( P S k h k 2 + N 0 ) + N 0 h g x S + g q P r (1 ) ( P S k h k 2 + N 0 ) + N 0 p 1 pn r + n p + n d (9) where n d is additiv e Gaussian noise with z ero mean and v ar iance N 0 and then, the signal-to-noise r atio (SNR) at the end-user can be calculated as . S N R = P S k h k 2 j g j 2 j g j 2 N 0 + j g j 2 N 0 1 + N 0 P r G ( ) 2 (1 ) (10) 3. OPTIMAL OUT A GE PERFORMANCE In this section, to address optimal tr ansmission mode , a R TS (requestto-send)/CTS (clear- to-send) scheme is deplo y ed with assumption of f aultless channel assessment. Bef ore star ting the inf or mation tr ansmission, the PB node sends a R TS pac k et to the the AP node . The AP node analyse receiv ed pilot-dr iv en data and can estimate the channel gain h. After receiving the R TS , the AP node retur ns a CTS pac k et to the source PB . As a result, the AP node can obtain full CSI, i.e ., h bef ore the entire tr ansmission. T o obtain the v alues of g at the PB , w e assume to use similar scheme . Similar ly , the AP can also estimate g b y itself and send f e edbac k to the PB node . It is w or th noting that the eff ect of channel estimation error is out of the scope of our paper . Through the R TS/CTS mechanism, which is compatib le with IEEE 802.11 ser ies standards , the AP and PB node ca n obtain the full CSI bef ore the entire tr ansmission po w er and data signal. S urely , e xtr a o v erhead and energy consumption are added to the R TS/CTS processing mechanism in ter m of channel estimation. But, this hard prob lem is out of topic of this paper In this in v estigation, the prob lem here is that using the kno wledge of h and g, w e aim to minimiz e the outage probability which corresponding to maximiz e the instantaneous SNR at the mobile user . It is w or th noting that the outage e v ent can be e xpressed as O P = Pr ( S N R < 0 ) ; (11) in which 0 is the threshold SNR. Theref ore , the optimization prob lem of outage probability can be wr itten as maximize S N R ( ) s:t: 0 < < 1 (12) T o der iv e optimal v alue of po w er splitting fr action of EH, it is noted that S N R ( ) is a nonnegativ e contin uous function . In f act, S N R ( ) is a c o nca v e function on 0 < < 1 . Here , the first order der iv ativ e of S N R ( ) with respect to optimal po w er splitting fr action is der iv ed as result of belo w equation @ S N R ( ) @ = 0 ; 0 < < 1 (13) Replacing e xpression of P r and G , w e obt ain ne w e xpression of the end-to-end SNR a s belo w S N R = P S k h k 2 j g j 2 (1 ) j g j 2 N 0 2 + 2 N 0 j g j 2 N 0 + N 0 + N 2 0 k h k 2 P S + N 0 (14) F or simplicity , it can be e xpressed SNR as S N R = ax (1 ) b 2 + c + d (15) Optimal Energy Har v esting Scheme f or P o w er Beacon-Assisted ... (Dinh-Thuan Do) Evaluation Warning : The document was created with Spire.PDF for Python.
806 ISSN: 2502-4752 in which, w e denote a = P S k h k 2 j g j 2 ; (16) b = j g j 2 N 0 ; (17) c = 2 N 0 j g j 2 N 0 ; (18) d = N 0 + N 2 0 = k h k 2 P S + N 0 (19) It is w or th noting that the der iv ation of SNR is too complicated as belo w @ ( S N R ) @ = a d 2 d + ( b c ) 2 ( d + ( c b )) 2 (20) T o obtain e xtreme v alues , w e need to solv e the equation @ ( ) @ = 0 . w e can see the sign of @ ( ) @ corresponds with the n umer ator ter m, f ( ) = d 2 dx + ( b c ) x 2 Hence w e just need to solv e f ( ) = 0 . In this proposed model, is f ound and then it can achie v e the maxim um SNR = ( 1 2 ; if b c = 0 d p bd + cd + d 2 b c ; if b c 6 = 0 (21) 4. NUMERICAL RESUL TS In this section, the analytical e xpressions are der iv ed to v alidate in the pre vious sections b y deplo ying Monte Car lo sim ulation results . All the sim ulation results are acquired b y a v er aging o v er 10 6 independent e xper iments . W e set the fix ed tr ansmission r ate R = 3 bps=H z , hence the outage SNR threshold is giv en b y = 2 R 1 = 7 . The energy har v esting efficiency is set to be = 0 : 9 . Also , w e set a v er age channel gain as h = g = 1 : 5 . W e obser v ed that the outage probability achie v ed b y the proposed po w er splitting policies as sho wn in Fig. 2. As can be seen clear ly , the a v er age outage probability decreases when the tr ansmit po w er increases . It can be confir med that the proposed po w er splitting policies attain better perf or mances than all the other patter ns with fix ed . That is to sa y the policy with par tial CSI, which incurs less o v erhead, approaches the policy with full CSI closely . If the R TS/CTS mechanism is not a v ailab le , policy with par tial CSI is suggested In Fig. 3, it can be freely obser v ed that dr iving more antennas at the PB can significantly increase the achie v ab le outage probability . This occurrence is quite intuitiv e , since increasing the n umber of antennas can pro vide higher energy beamf or ming gain, hence , the amount of the har v ested energy at the source impro v es , which in tur n reduces the outage probability of the system. This confir ms the role of PB in f eeding energy to wireless node . 5. CONCLUSION This paper considered a point-to-point wireless po w ered comm unication system, which ma y find potential applications in future netw or ks such as medical, sensor , and underw ater com- m unications systems . A detailed in v estigation on the a v er age outage perf or mance of such sys- tems w as presented with optimal po w er splitting fr action of nergy har v esting protocol. F or po w er beacon equippe d m ulti an tenna f or energy tr ansf er tr ansmission modes , the n umber of antenna contr ib ute to better the a v er age outage perf or mance as seen in n umer ical result. In addition, a solution of the optimal splitting fr action to minimiz e the a v er age ouatge w as e xamined, and sim- ulation w ere obtained, which w ere sho wn to be v er y accur ate . Since the optimal split depend on the instantaneous channel state in f or mation, it is a lo w comple xity solution to enhance the system perf or mance . IJEECS V ol. 7, No . 3, September 2017 : 802 808 Evaluation Warning : The document was created with Spire.PDF for Python.
IJEECS ISSN: 2502-4752 807 0 5 10 15 20 25 30 35 40 10 −6 10 −5 10 −4 10 −3 10 −2 10 −1 10 0 P S (dBm) P o u t     optimal  ρ fixed  ρ =0.4 fixed  ρ =0.6 Figure 2. Outage probability at destination node with respect to energy har v esting po w er splitting fr actions 0 5 10 15 20 25 30 35 40 10 −6 10 −5 10 −4 10 −3 10 −2 10 −1 10 0 P S (dBm) P o u t     number of antenna = 2 number of antenna = 4 number of antenna = 6 Figure 3. Outage probability destination node with respect to n umber of antenna at PB Ref erences [1] W . Lumpkins , Nik ola T eslas dream realiz ed: Wireless po w er energy har v esting, IEEE Con- sumer Electron. Mag., v ol. 3, no . 1, pp . 3942, J an. 2014. [2] B . Medepally and N. B . Mehta, V oluntar y energy har v esting rela ys and selection in cooper ativ e wireless netw or ks , IEEE T r ans . Wireless Comm un., v ol. 9, no . 11, pp . 3543–3553, No v . 2010. [3] Dinh-Thuan Do , ”Optimal Throughput under Time P o w er Switching based Rela ying Protocol in Optimal Energy Har v esting Scheme f or P o w er Beacon-Assisted ... (Dinh-Thuan Do) Evaluation Warning : The document was created with Spire.PDF for Python.
808 ISSN: 2502-4752 Energy Har v esting Cooper ativ e Netw or k”, Wireless P ersonal Comm unications (Spr inger), V ol. 87, No . 2, pp . 551–564, 2016 [4] Dinh-Thuan Do , ”Energy-A w are T w o-W a y Rela ying Netw or ks under Imperf ect Hardw are: Op- timal Throughput Design and Analysis”, T elecomm unication S y st ems Jour nal (Spr inger), V ol. 62, No . 2, pp . 449–459, 2015 [5] A. Minasian, S . ShahbazP anahi, and R. S . Adv e , ”Energy har v esting cooper ativ e comm unica- tion systems , IEEE T r ans . Wireless Comm un., v ol. 13, no . 11, pp . 6118–6131, No v . 2014 [6] I. Ahmed, A. Ikhlef , R. Schober , and R. K. Ma llik, ”Joint po w er allocation and re la y selection in energy har v esting AF rela y systems , IEEE Wi reless Comm un. Lett., v ol. 2, no . 2, pp . 239–242, Apr . 2013. [7] I. Ahmed, A. Ikhlef , R. Schober , and R. K. Mallik, ”P o w er allocation in energy har v esting rela y systems , in Proc. IEEE V eh. T echnol. Conf . (VTC), 2012, pp . 1–5. [8] I. Kr ikidis , T . Char ala mbous , and T . S . Thompson, ”Stability analysis and po w er allocation f or energy har v esting cooper ativ e netw or ks , IEEE Signal Process . Lett., v ol. 19, no . 1, pp . 20–23, J an. 2012. [9] I. Ahmed, A. Ikhlef , R. Schober , and R. K. Mallik, ”P o w er allocation f or con v entional and b uff er- aided link adaptiv e rela ying systems with energy har v esting nodes , IEEE T r ans . Wireless Com- m un., v ol. 13, no . 3, pp . 1182–1195, Mar . 2014. [10] K. Huang and V . Lau, ”Enab ling wireless po w er tr ansf er in cellular netw or ks: architecture , modeling and deplo yment, IEEE T r ans . Wireless Comm un., v ol. 13, no . 2, pp . 902–912, F eb . 2014 IJEECS V ol. 7, No . 3, September 2017 : 802 808 Evaluation Warning : The document was created with Spire.PDF for Python.