Energy Harvesting Based Body Area Networks for Smart Health

Transcription

1 sesors Article Eergy Harvestig Based Body Area Networks for Smart Health Yixue Hao 1, Limei Peg 2, *, Huimi Lu 3, Mohammad Mehedi Hassa 4 ad Atif Alamri 4 1 School of Computer Sciece ad Techology, Huazhog Uiversity of Sciece ad Techology, Wuha , Chia; yixuehao@hust.edu.c 2 Departmet of Idustrial ad Iformatio System Egieerig, Ajou Uiversity, Suwo , Korea 3 Departmet of Mechaical ad Cotrol Egieerig, Kyushu Istitute of Techology, Fukuoka prefecture , Japa; luhuimi@ieee.org 4 College of Computer ad Iformatio Scieces, Kig Saud Uiversity, Riyadh 11543, Saudi Arabia; mmhassa@ksu.edu.sa (M.M.H.); atif@ksu.edu.sa (A.A.) * Correspodece: auroraplm@ajou.ac.kr; Tel.: Received: 7 May 2017; Accepted: 1 July 2017; Published: 10 July 2017 Abstract: Body area etworks (BANs) are cofigured with a great umber of ultra-low power cosumptio wearable devices, which costatly moitor physiological sigals of the huma body ad thus realize itelliget moitorig. However, the collectio ad trasfer of huma body sigals cosume eergy, ad cosiderig the comfort demad of wearable devices, both the size ad the capacity of a wearable device s battery are limited. Thus, miimizig the eergy cosumptio of wearable devices ad optimizig the BAN eergy efficiecy is still a challegig problem. Therefore, i this paper, we propose a eergy harvestig-based BAN for smart health ad discuss a optimal resource allocatio scheme to improve BAN eergy efficiecy. Specifically, firstly, cosiderig eergy harvestig i a BAN ad the time limits of huma body sigal trasfer, we formulate the eergy efficiecy optimizatio problem of time divisio for wireless eergy trasfer ad wireless iformatio trasfer. Secodly, we covert the optimizatio problem ito a covex optimizatio problem uder a liear costrait ad propose a closed-form solutio to the problem. Fially, simulatio results proved that whe the size of data acquired by the wearable devices is small, the proportio of eergy cosumed by the circuit ad sigal acquisitio of the wearable devices is big, ad whe the size of data acquired by the wearable devices is big, the eergy cosumed by the sigal trasfer of the wearable device is decisive. Keywords: eergy harvestig; wireless powered body area etwork; resource allocatio 1. Itroductio Body area etworks (BANs) are small wireless sesor etworks (WSNs) which support a lot of medical applicatios ad provide a solutio for smart health moitorig [1]. For a exhaustive itroductio to BANs, we refer the reader to [2,3]. BANs are cofigured with ultra-low power cosumptio wearable devices [4] ad medical sesors [5] (such as digestible medical electroics). These sesors costatly moitor physiological sigals ad movemet data of the huma body; they trasfer such sigals ad data to the cloud for aalysis, thus realizig itelliget moitorig of the user s health [6,7]. However, data collectio ad trasfer of BAN sesors cosume eergy. To icrease the comfort of wearable devices, the battery of wearable devices is usually small, thus the battery s capacity is limited [8,9]. Old batteries eed to be replaced frequetly or recharged regularly. For medical sesors (such as digestible sesors), it is impossible to replace or recharge the battery. Thus, miimizig the Sesors 2017, 17, 1602; doi: /s

2 Sesors 2017, 17, of 10 eergy cosumptio of wearable devices ad optimizig BAN eergy efficiecy is still a challegig problem [10]. Nowadays, most research cocerig BAN eergy efficiecy optimizatio focuses o the desig of the routig algorithm [11,12], duty-cycle-based data collatio [1], data reductio ad compressed sedig [13,14], ad cross-layer desig [15]. For example, some papers have proposed ovel approaches to reduce eergy cosumptio through a adaptive routig algorithm [16], dyamic programmig for heterogeeous etworks [17], ad voltage/frequecy scalig [18], ad secure data trasmissio [19]. Other papers have dealt with the issue of data-geeratio ucertaity i the optimal desig of BANs: [20,21] proposed a robust optimizatio model solved by fast mixed iteger programmig heuristics, based o the algorithm for robust capacitated etwork desig proposed i [22]; Referece [23] has istead ivestigated the adoptio of a mi max regret model. However, few works realize BAN eergy efficiecy optimizatio through eergy harvestig. Geerally, eergy i the BAN sesor ca be harvested i the followig three ways: Eergy harvestig through the eviromet: some sesors harvest eergy through reewable eergy sources (icludig solar, wid, ad lumious eergy resources). Eergy harvestig through the huma body: some sesors harvest eergy from their ow heat eergy, bio-eergy, body surface frictio ad body movemet. Eergy harvestig through a wireless sigal: some sesors harvest eergy by acquirig wireless sigals. For eergy harvestig through the eviromet, it is impossible for the user to be exposed to strog sushie or strog wid for log time ad this eergy harvestig mode depeds greatly o the weather ad other coditios, which may lead to a loger delay [24,25]. Thus, it is ot applicable to BANs. For eergy harvestig through the huma body, the bio-eergy of the huma body is ustable ad may result i ureliable eergy productio; furthermore, wearig a additioal eergy-harvestig device may result i discomfort [26]. Thus, it is also ot applicable to a BAN. For eergy harvestig through a wireless sigal or radio frequecy (RF) eergy harvestig [27,28], cosiderig that wireless sigals exist everywhere costatly ad cotrollably, it is a feasible approach to provide reliable eergy to the low power cosumptio sesors. A RF eergy harvestig based BAN icludes two stages: wireless eergy trasfer (WET) ad wireless iformatio trasfer (WIT) [29,30]. A challegig problem of the simultaeous wireless iformatio ad power trasfer (SWIPT) is how to allocate resources betwee the WET ad WIT, so as to miimize the eergy cosumed by the etwork. Some research has icluded primary exploratio ad discussio o BANs. For example, Abubaka et al. [5] proved i their research that WET to the sesor i the digestive tract could be realized through a atea outside the huma body ad such eergy was sufficiet to keep the sesor workig ormally, icludig beig able to moitor the eviromet ad temperature of the digestive tract ad the special utritio cost.however, these studies failed to take ito cosideratio the eergy cosumed by the circuit ad data collectio ad processig of the sesor. I fact, the small size of the sesor i a BAN may result i a great proportio of eergy beig cosumed by the circuit, data collectio ad processig. Thus, it is obviously impossible to eglect such power cosumptio. I this paper, we propose the eergy harvestig-based BAN, i.e., the sesor i a BAN ca harvest eergy from access poits ad trasmit the collected data to access poits. Furthermore, we study the resources allocatio scheme that realizes miimized eergy cosumptio i a BAN. To be specific, the mai results ad cotributios of this paper iclude the followig: We itroduce eergy harvestig ito a BAN to improve BAN eergy efficiecy. Compared with traditioal BANs, the eergy harvestig based BAN proposed i this paper ca sigificatly improve the BAN eergy efficiecy. We formulate the optimizatio problem cocerig time allocatio for the WET ad WIT i a BAN, with the aim of miimizig eergy cosumptio i the sesor whe cosiderig the

3 Sesors 2017, 17, of 10 WIT ad WET time limits. Furthermore, we covert such a problem ito a covex optimizatio problem uder liear costraits. We propose a closed-form solutio to the optimizatio problem based o Karush Kuh Tucker (KKT) coditios. Simulatio results showed that whe the size of data acquired by the wearable devices is small, the proportio of eergy cosumed by the circuit ad iformatio collectio of the wearable devices is big, ad whe the size of data acquired by the wearable devices is big, the eergy cosumed by iformatio trasfer of the wearable device is decisive. The remaider of this article is orgaized as follows. The system model is described i Sectio 2. We first formulate a optimizatio problem to miimize the eergy cosumptio of the eergy harvestig-based BAN. The, we trasform the problem to a covex optimizatio problem with liear costraits ad propose a closed-form solutio i Sectio 3. Our experimetal results ad discussios are provided i Sectio 4. Fially, Sectio 5 cocludes this paper. 2. System Model I this sectio, we itroduce the etwork architecture of a eergy harvestig-based BAN, as well as the WET ad WIT model based o the time divisio multiple access (TDMA) protocol Eergy Harvestig-Based Body Area Networks Model We cosider a eergy harvestig BAN as show i Figure 1; the ultra-low power cosumptio sesor cofigured to the BAN could collect the physiological sigals of the huma body, such as electrocardiography sigals. The access poit recharged the sesor through WET at a fixed iterval ad the sesor eeded to deliver collected physiological sigals of the huma body to the access poit. That is, the huma body sesor harvested eergy from the access poit. Whe the sesor acquired eergy, some of the harvested eergy was used for sigal trasfer, while some of the harvested eergy was cosumed by the circuit ad data acquisitio of the sesor. I this paper, we assume that the access poit has a stable eergy supply ad ca provide sufficiet eergy to the sesor. I this paper, the trasfer protocol used is the TDMA protocol as show i Figure 2. We assume that there are sesors withi the area covered by the access poit ad these sesors are deoted as S = {S 1, S 2,, S }. Let t 0, t 1,, t deote the time slot ad T t0, T t1,, T t represet the duratio of the time slot. Cosiderig that the WET ad WIT will be fiished withi the period T, T ca be divided ito + 1 duratio of the time slot, i.e., T ti = T. (1) i=0 I the duratio of the time slot t 0, WET is performed ad the access poit trasfers eergy to sesors by way of sigal broadcastig. I the duratio of the time slot t 1, t 2,, t, sesors perform WIT ad trasfer iformatio to the access poit through the harvested eergy. Wearable Devices Access Poit Body Area Networks Figure 1. Eergy harvestig-based body area etwork.

4 Sesors 2017, 17, of 10 T Wireless Eergy Trasfer Wireless Iformatio Trasfer Tt 0 t 0 T-Tt 0 S 1... S 2 S -1 S 2.2. Trasmissio Model Tt 1 Tt 2 Tt t 0 t 1 t 2 t -1 Tt -2 t-1 t Figure 2. The time divisio multiple access (TDMA)-based trasmissio protocol. The trasmissio modes icluded WET ad WIT. I WET, the sesor harvests eergy through the access poit. I WET, the sesor trasfers acquired iformatio to the access poit. Detailed models are show below. Wireless eergy trasfer model: Let P b deote the access poit trasmissio power; the, accordig to the the work of You et al. [31], the eergy harvested by the sesor S i, deoted as E H i, is give as: E H i = ηp b h DL i T t0 (2) where hi DL is the chael gais from the access poit to the sesor S i, ad η (0 < η < 1) is the eergy covertig efficiecy. Iformatio trasfer model: After the sesor harvested eergy, it is ecessary to trasmit the collected iformatio to the access poit. The chael gai of the sesor S i is defied as, ad the trasfer power of S i is defied as p i. The, accordig to the work of You et al. [31], the trasfer rate of S i i the time slot t i, deoted as r i, is give as: ( r i = B log p ) i σ 2 where σ 2 is the variace of complex white Gaussia oise, B is the chael badwidth from the sesor S i to the access poit Eergy Cosumptio Model The eergy cosumptio of the sesor S i i T ti ca be divided ito the followig three parts: 1. Eergy cosumed by the circuit of the sesor S i. Cosiderig that the sesor works costatly, this eergy is costat ad we deote it as E c i. 2. Eergy cosumed by sigals (such as the perceptio, collectio ad storage of sigals) processed by the sesor S i. This part of eergy cosumptio is associated with the data size whics deoted as to be processed. Let γ i deote the eergy cosumed by the processig of oe bit of data. The, this part of eergy ca be expressed as E proc i = γ i. 3. Eergy cosumed for iformatio trasfer by the sesor S i. This part of eergy cosumptio is also related to the data size ad ca be represeted as E tra i = p i /r i. Accordig to the work of You et al. [31], uder a give time costrait, the most eergy-efficiet data trasfer policy is fixed-rate trasmissio over the whole time slot. Thus, for the sesor S i i the duratio of time slot t i, the lowest eergy cosumptio trasfer rate is fixed at r i = /T ti. Based o the above discussio ad Equatio (5), we ca rewrite the Ei tra as follows: E tra i = p i T ti = (2 (3) 1) σ2 T ti (4)

5 Sesors 2017, 17, of 10 Thus, the total eergy cosumed by the sesor S i, deoted as Ei loc, ca be obtaied as follows: E loc i = E c i + Eproc i 3. Problem Formulatio ad Solutio + Ei tra = Ei c ω + γ i Bt i + (2 i 1) σ2 t i (5) I this sectio, we itroduce the optimizatio problem of BAN time allocatio, with the aim of achievig ratioal allocatio of resources, thus miimizig the eergy cosumptio of the sesor Problem Formulatio I this paper, we assume that T t0 is fixed. That is, the system gives the parameter T t0 earlier, i which eergy is trasferred to sesors i the huma body. T opt = [T t1, T t2,, T t ] with the aim of dividig the period T T t0 ratioally, thus miimizig the eergy cosumptio of the sesor. I view of the discussio described i Sectio II, the followig optimizatio problems ca be obtaied: P1 : miimize T ti subject to: [ E c i + γ i + (2 1) σ2 T ti ] T ti T T t0, T ti 0, i = 1, 2,,. (7) E c i + γ i + (2 1) σ2 T ti (6) η i P b h DL i T t0, i = 1, 2,,. (8) where the objective fuctio (6) is the eergy cosumptio of the miimized sesor. The costrait coditio (7) meat that the WIT of sesors was fiished withi the time duratio T T t0.the costrait coditio (8) meat that the eergy cosumed by the sesor should ot exceed the harvested eergy. To solve the above optimizatio problem, we further adapt the problem ad combie the costrait coditio (8) ad the objective fuctio (6); the, we ca obtai the eergy savigs E i as follows: [ ] E i = η i P b hi DL T t0 E c i + γ i + (2 1) σ2 T ti Thus, the followig optimizatio problem, whics equivalet to P1, ca be obtaied: (9) P2 : miimize t i subject to: E i (10) t i T T t0, T ti 0, i = 1, 2,,. (11) I the ext subsectio, we characterize the solutio of the problem Closed-Form Solutio As for the optimizatio problem described above, it ca be prove to be a covex optimizatio problem as follows. Theorem 1. Problem P2 is the covex optimizatio problem.

6 Sesors 2017, 17, of 10 Proof of Theorem 1. Defie the fuctio f (x) = 2 x B 1; the, we ca obtai the first ad the secod derivative of f (x) as follows: f (x) = l 2 B 2 x B, f (x) = ( ) l B x B Here, it is clear that f (x) > 0, thus f (x) is covex. Cosiderig the perspective fuctio of f (x), g(x, t) = t f (x/t) is also covex with respect to (t, x). It is obvious that Ei t = t i (2 /t i h 2 B 1)σ 2 is k the covex fuctio of t i. Therefore, Ei c + Eproc i + Ei tra is a covex fuctio. Sice the sum of covex fuctios is still covex, the objective fuctio is a covex fuctio. The restrictio is a liear costrait. Thus, the optimizatio problem is covex [32]. As for problem P2, we ca defie its Lagrage fuctio as follows: L = [ Ei c + γ i + (2 1) σ2 T ti η i P b hi DL T t0 ] + λ ( T ti + T t0 T ) (12) where λ is the Lagrage multiplier. It is assumed that T t i ad λ are the optimal solutios of problem P2 ad its dual problem, respectively. Based o the Karush Kuh Tucker (KTT) coditios, the followig coditios ca be obtaied: T ti T T t0 (13) λ 0 ) (14) λ ( T t i + T t0 T L Tt = 2 i σ 2 (1 l 2 = 0 (15) ) σ2 + λ = 0 (16) Based o these coditios, the optimal time allocatio scheme ca be derived as the followig theorem. Theorem 2. The optimal time allocatio scheme for Problem P2 is show as follows. T t i = l 2 B(1 + W( λ σ 2 eσ 2 )) (17) where W(x) is the Lambert fuctio ad T t i = T T t0. Proof of Theorem 2. For the sake of simplicity, we defie A = σ2, x = BT t i. Based o (16), it has: 2 x (1 x l 2) = A λ A (18) Usig the properties of expoetial fuctios, we ca obtai: e x l 2 = A l 2 λ A (x 1 l 2 ) (19)

7 Sesors 2017, 17, of 10 Based o the defiitio of the Lambert fuctio, the solutio of (19) ca be obtaied: x = 1 + W( λ σ2 eσ 2 ) l 2 (20) Thus, the optimal resource allocatio scheme will be give as follows: T t i = l 2 B(1 + W( λ σ 2 eσ 2 )) (21) Furthermore, based o (15), we ca obtai λ = 0 or T t i + T t0 T = 0. (1) If λ = 0, the optimal resource allocatio scheme would be show as below: T t i = l 2 B(1 + W( e 1 )) (22) sice W( e 1 ) = 1, the deomiator of (22) is zero. Thus, λ satisfy λ > 0. (2) If λ > 0, the optimal resource allocatio scheme would be obtaied as follows: T t i = l 2 B(1 + W( λ σ 2 eσ 2 )) (23) ad it satisfies T t i = T T t0. 4. Simulatio Results I this sectio, we coduct the simulatio experimets. Firstly, we set the simulatio parameter, the we evaluate the performace Parameter Settig We assume that there are 6 sesors i BAN. As for the WET stage, assume that T = 1 s ad the WET time slot is t 0 = 200 ms. The chaels are modeled as idepedet Rayleigh fadig with average power loss set as We set the trasmitted power of the access poit as P b = 100 W. For the WIT stage, we set the badwidth as W = 5 MHz ad the variace of complex white Gaussia chael oise as σ = 10 9 W. I this paper, for the purpose of coveiece, we assume that the sesors have equal circuit cosumptio E c i ad eergy γ i cosumed for processig oe bit of data. Hece, i this paper, E i = J ad γ = 10 4 J/bit. The data size to be trasferred followed uiform distributio with the mea value ω = 1000 bit Eergy Cost of Sesors We give the effects of data size ω o the eergy cosumptio of the sesor. The X-axis represets the size of data (Kbits), the Y-axis is the eergy cosumptio of all the sesors ad we used logarithmic coordiates to the axis Y, ad the correspodig uit of measuremet is log 10 Joule. It ca be see from Figure 3 that alog with the icrease i data size ω, more eergy will be cosumed. It is because the icrease of data size requires more eergy for data trasfer ad iformatio processig. It ca be observed from Figure 3a that with the same size of data trasfer, more eergy cosumed i processig each bit of data suggests more eergy cosumed by the system. It is clear i Figure 3b that with the same trasferred data size, the bigger the eergy E c cosumed by the circuit of the sesor, the bigger the eergy cosumed by the sesor. Furthermore, it ca be deduced based o these two figures that whe the trasfer data size is smaller, for istace, ω = 1000 bits, the differece betwee curves is bigger. It is because whe s smaller, the eergy cosumed by the circuit of the sesor ad the eergy cosumed by iformatio processig take a greater proportio; whe the

8 Sesors 2017, 17, of 10 trasfer data size is bigger, such as ω = 1200 bits, the differece betwee curves decreases because whe the trasfer data size s big, the eergy cosumptio of the sesor is maily eergy cosumed for data trasfer. Moreover, Figure 4 shows that more eergy is required as the umber of sesors icreases. The reaso is that as the umber of sesors icreases, sesors ca collect more data ad deliver the data to the access poit, which cosumes more eergy Eergy cost (log 10 (J)) γ= γ= γ= γ= Eergy cost (log 10 (J)) E c =0.001J E c =0.008J E c =0.016J E c =0.024J Trasmissio data (Kbits) (a) Trasmissio data (Kbits) (b) Figure 3. (a) Effects of trasmissio data size ω o eergy cosumptio uder differet γ; (b) Effects of trasmissio data size ω o eergy cosumptio uder differet E c. Eergy cost (log 10 (J)) =4 =5 =6 = Trasmissio data (Kbits) Figure 4. Effects of trasmissio data size ω o eergy cosumptio uder differet Time Duratio Allocatio of Sesors I this subsectio, we discuss the relatioship betwee trasmissio data size ω ad time duratio allocatio T ti. Figure 5a shows the effects of trasmissio data size ω o the time duratio allocatio. The X-axis represets the mea value of trasmissio data size that follows a uiform distributio. The Y-axis is the time duratio allocatio. From the figure, we ca observe that the time duratio allocated by sesors 1, 2 ad 3 varies little with the icremet of trasmissio data size. Compared with Figure 3, we ca coclude that the size of data trasmitted has a greater impact o eergy cost tha o the time duratio allocatio. Figure 5b shows how the system allocates time duratio for each sesor whe give a set of trasmissio data sizes. The X-axis represets the specific trasmissio data size ad the Y-axis idicates the time duratio allocatio. From the figure, we ca observe that the time duratio allocatio icreases with the icremet of give trasmissio data.

9 Sesors 2017, 17, of 10 Time duratio allocatio (ms) sesor1 sesor2 sesor3 Time duratio allocatio (ms) Trasmissio data (Kbits) (a) Trasmissio data (Kbits) (b) Figure 5. (a) Effects of trasmissio data size ω o time duratio allocatio uder differet sesors; (b) Effects of trasmissio data size ω o time duratio allocatio uder a give data size. 5. Coclusios I this paper, we proposed a eergy harvestig-based body sesor etwork, ad a method based o time divisio multiple access (TDMA); we built upo the optimizatio problem of time divisio for wireless iformatio trasfer ad proved that this optimizatio problem was actually a covex optimizatio problem; we gave a closed-form solutio to the problem. The simulatio results of the experimet idicated that whe the size of data acquired by the sesor was small, the eergy cosumptio of the sesor was maily eergy cosumed by the sesor circuit ad eergy cosumed for data acquisitio; whe the size of data acquired by the sesor was big, the eergy cosumed by the sesor for data trasfer was decisive. Ackowledgmets: The authors would like to exted their appreciatio to the Deaship of Scietific Research at Kig Saud Uiversity for fudig this work through research group No. RGP-281. Author Cotributios: Yixue Hao ad Limei Peg defied the research theme; Huimi Lu developed the mathematical model; Mohammad Mehedi Hassa adatif Alamri made critical revisio of the article. All authors have cotributed to the productio of the paper ad have approved the mauscript Coflicts of Iterest: The authors declare o coflict of iterest. Refereces 1. Habib, C.; Makhoul, A.; Darazi, R.; Salim, C. Self-adaptive data collectio ad fusio for health moitorig based o body sesor etworks. IEEE Tras. Id. Iform. 2016, 12, Che, M.; Gozalez, S.; Vasilakos, A.; Cao, H.; Leug, V.C. Body area etworks: A survey. Mob. Netw. Appl. 2011, 16, Negra, R.; Jemili, I.; Belghith, A. Wireless Body Area Networks: Applicatios ad Techologies. Procedia Comput. Sci. 2016, 83, Che, M.; Ma, Y.; Li, Y.; Wu, D.; Zhag, Y.; You, C.H. Wearable 2.0: Eablig Huma-Cloud Itegratio i Next Geeratio Healthcare Systems. IEEE Commu. Mag. 2017, 55, Abid, A.; O Brie, J.M.; Besel, T.; Clevelad, C.; Booth, L.; Smith, B.R.; Lager, R.; Traverso, G. Wireless Power Trasfer to Millimeter-Sized Gastroitestial Electroics Validated i a Swie Model. Sci. Rep. 2017, 7, Che, M.; Hao, Y.; Hwag, K.; Wag, L.; Wag, L. Disease Predictio by Machie Learig over Big Data from Healthcare Commuities. IEEE Access 2017, 5, Che, M.; Zhou, P.; Fortio, G. Emotio Commuicatio System. IEEE Access 2017, 5, Li, K.; Sog, J.; Luo, J.; Ji, W.; Hossai, M.S.; Ghoeim, A. GVT: Gree video trasmissio i the mobile cloud etworks. IEEE Tras. Circuits Syst. Video Techol. 2017, 27, Li, K.; Luo, J.; Hu, L.; Hossai, M.S.; Ghoeim, A. Localizatio based o Social Big Data Aalysis i the Vehicular Networks. IEEE Tras. Id. Iform. 2016, doi: /tii

10 Sesors 2017, 17, of Che, M.; Yag, J.; Hao, Y.; Mao, S.; Hwag, K. A 5G cogitive system for healthcare. Big Data Cog. Comput. 2017, 1, Zhou, Y.; Sheg, Z.; Mahapatra, C.; Leug, V.C.; Servati, P. Topology desig ad cross-layer optimizatio for wireless body sesor etworks. Ad Hoc Netw. 2017, 59, Tsouri, G.R.; Prieto, A.; Argade, N. O icreasig etwork lifetime i body area etworks usig global routig with eergy cosumptio balacig. Sesors 2012, 12, Gravia, R.; Aliia, P.; Ghasemzadeh, H.; Fortio, G. Multi-sesor fusio i body sesor etworks: State-of-the-art ad research challeges. If. Fusio 2017, 35, Zhag, H.; Liu, J.; Pag, A.C.; Li, R. A Data Recostructio Model Addressig Loss ad Faults i Medical Body Sesor Networks. I Proceedigs of the IEEE Global Commuicatios Coferece (GLOBECOM), Washigto, DC, USA, 4 8 December Che, X.; Xu, Y.; Liu, A. Cross Layer Desig for Optimizig Trasmissio Reliability, Eergy Efficiecy, ad Lifetime i Body Sesor Networks. Sesors 2017, 17, Qiu, M.; Mig, Z.; Li, J.; Liu, J.; Qua, G.; Zhu, Y. Iformer homed routig fault tolerace mechaism for wireless sesor etworks. J. Syst. Archit. 2013, 59, Qiu, M.; Sha, E.H.M. Cost miimizatio while satisfyig hard/soft timig costraits for heterogeeous embedded systems. ACM Tras. Des. Autom. Electro. Syst. 2009, 14, Qiu, M.; Mig, Z.; Li, J.; Liu, S.; Wag, B.; Lu, Z. Three-phase time-aware eergy miimizatio with DVFS ad urollig for chip multiprocessors. J. Syst. Archit. 2012, 58, Zhou, L.; Wu, D.; Zheg, B.; Guizai, M. Joit physical-applicatio layer security for wireless multimedia delivery. IEEE Commu. Mag. 2014, 52, D Adreagiovai, F.; Nardi, A. Towards the fast ad robust optimal desig of Wireless Body Area Networks. Appl. Soft Comput. 2015, 37, D Adreagiovai, F.; Nardi, A.; Natalizio, E. A fast ILP-based Heuristic for the robust desig of Body Wireless Sesor Networks. I Applicatios of Evolutioary Computatio; Spriger: Viea, Austria, 2017; Volume 10199, pp D Adreagiovai, F.; Krolikowski, J.; Pulaj, J. A fast hybrid primal heuristic for multibad robust capacitated etwork desig with multiple time periods. Appl. Soft Comput. 2015, 26, D Adreagiovai, F.; Nace, D.; Nardi, A.; Natalizio, E. Robust relay ode placemet i body area etworks by heuristic mi-max regret. I Proceedigs of the IEEE Balka Coferece o Commuicatios ad Networkig (BALKANCOM), Tiraa, Albaia, 30 May 2 Jue Li, K.; Che, M.; Deg, J.; Hassa, M.M.; Fortio, G. Ehaced figerpritig ad trajectory predictio for IoT localizatio i smart buildigs. IEEE Tras. Autom. Sci. Eg. 2016, 13, Zhou, L. QoE-drive delay aoucemet for cloud mobile media. IEEE Tras. Circuits Syst. Video Techol. 2017, 27, Che, M.; Ma, Y.; Sog, J.; Lai, C.F.; Hu, B. Smart clothig: Coectig huma with clouds ad big data for sustaiable health moitorig. Mob. Netw. Appl. 2016, 21, Ghazafari, A.; Tabassum, H.; Hossai, E. Ambiet RF eergy harvestig i ultra-dese small cell etworks: Performace ad trade-offs. IEEE Wirel. Commu. 2016, 23, Hao, Y.; Che, M.; Hu, L.; Sog, J.; Volk, M.; Humar, I. Wireless Fractal Ultra-Dese Cellular Networks. Sesors 2017, 17, Xiog, K.; Fa, P.; Zhag, C.; Letaief, K.B. Wireless iformatio ad eergy trasfer for two-hop o-regeerative MIMO-OFDM relay etworks. IEEE J. Sel. Areas Commu. 2015, 33, Bi, S.; Zhag, R. Placemet optimizatio of eergy ad iformatio access poits i wireless powered commuicatio etworks. IEEE Tras. Wirel. Commu. 2016, 15, You, C.; Huag, K.; Chae, H. Eergy efficiet mobile cloud computig powered by wireless eergy trasfer. IEEE J. Sel. Areas Commu. 2016, 34, Boyd, S.; Vadeberghe, L. Covex Optimizatio; Cambridge Uiversity Press: Cambrige, UK, c 2017 by the authors. Licesee MDPI, Basel, Switzerlad. This article is a ope access article distributed uder the terms ad coditios of the Creative Commos Attributio (CC BY) licese (