A Three-Dimensional Network Coverage Optimization Algorithm in Healthcare System

Transcription

1 204 IEEE 6th Internatonal Conference on e-health Networkng, Applcatons and Servces (Healthcom) A Three-Dmensonal Network Coverage Optmzaton Algorthm n Healthcare System Xaoshuang Lu, Guxa Kang, Nngbo Zhang, Bngnng zhu, Congcong L, Y Cha, Yuncheng Lu 2.Key Laboratory of Unversal Wreless Communcaton, Mnstry of Educaton Beng Unversty of Posts and Telecommuncatons Emals: LXS_55@6.com, gxkang@bupt.edu.cn 2.TGLD Informaton Centre Beng, Chna Abstract Ths paper presents a healthcare montorng archtecture coupled wth a wreless sensor network and wearable sensor systems whch montor chronc patents n nursng house or the elderly n ther home. Wth ths archtecture, we nvestgate how sensor nodes are deployed n the three-dmenson (D) montorng regon to acheve unform dstrbuton, whch drectly determnes the Qualty of Servce (QoS). Based on the exstng two-dmenson (2D) coverage-enhancng algorthms for wreless sensor networks, a D sensng model and a coverage optmzaton algorthm are proposed n ths paper. Frstly, an ntellgent optmzaton algorthm s utlzed to adust the poston of sensor nodes. Then, we pck out the redundant nodes wth the set coverage algorthm, and move them nto the uncovered area to ncrease the coverage rato. The smulaton results show that the coverage rato ncreased by the coverage optmzaton algorthm compared wth the other coverage algorthms. Keywords Healthcare, three-dmenson sensng model, coverage, ntellgent optmzaton, redundant nodes. I. INTRODUCTION Wreless sensor networks (WSNs) are wdely used n ndustry, agrculture, mltary, medcal care, and safety montorng []. Moble, wreless, pervasve computng and communcaton envronments are changng the way that medcal staffs nteract wth ther patents and the elderly. By deployng self-organzed sensor nodes for physologcal montorng reasonably, t s convenent to guarantee tmely nterventon by a physcan wth contnual patent montorng nformaton. Deployng nodes n nursng-house, home or dsaster area reasonably s mportant to the healthcare practtoners or emergency personnel. As a fundamental problem n WSNs, network coverage reflects the perceved performance of the network, whch can provde the QoS (Qualty of Servce). Tradtonal researches have focused on the two-dmenson (2D) network [2, ]. However, wth the demand for practcal applcaton, the deal 2D sensng model couldn t always be appled n the realstc envronment, such as home, hlls and the volcano [4]. Recently, the emergng feld n underwater coverage promoted the development of the research n threedmenson (D) WSN [5]. There are two types of deployment n D WSNs. One s Ths work s supported by Internatonal Scence and Technology Cooperaton Proect (no. 200DFA590), the Important Natonal Scence & Technology Specfc Proects of the Mnstry of Scence and Technology of Chna (no. 20ZX000600), and New Century Excellent Talents n Unversty (NCET) (no. NCET--059). determnstc deployment [6], and the other s random deployment. To obtan the expected coverage rato, we need to adust the nodes locatons or other parameters of the random deployment strateges [7]. The current studes are manly dvded nto two ways. One s puttng D problems nto the 2D plane [8], whch couldn t be solved effcently ust by the extenson of the 2D solutons. The other s usng the D mathematcal model and space geometrc theory [4] to solve the D coverage problems. In summary, some coverage optmzaton algorthm are based on graph theory and detecton algorthm. It s mpractcal to apply Graph theory n the montorng area where any pont s covered by a sensor node [9]. The detecton algorthm s ust sutable for small-scale WSNs snce t can t ensure a complete coverage [0]. Recently, wth the development of the heurstc search algorthms, we take advantage of ther propertes to solve the coverage problems, such as the fast searchng speed and strong processng. To solve the coverage rato, the researchers proposed the genetc algorthm (GA), partcle swarm optmzaton algorthm (PSOA), and ant colony algorthm (ACA) whch all mproved the overall performance to a certan extent, such as rasng the coverage rato or savng energy. However, for dfferent topologes of WSNs, t s rather complcated to adust the genetc operaton of GA []. PSOA [2] s easly trapped n local mnma value, whch s dffcult to converge to the global extreme value. Local search ablty of ACA s strong, but the solvng speed s too slow to operate ntally, whch affects the real-tme performance of the coverage optmzaton []. The drawbacks of GA, PSOA and ACA lmt ther applcatons n the network coverage optmzaton. In some cases, they even lead to a low coverage rato, the neffectve use of energy and a short lfetme. The artfcal fsh swarm algorthm (AFSA) [4] overcomes the defects of these algorthms, and we propose an optmal algorthm based on t n ths paper. The rest of the paper s structured as follows. A healthcare montorng archtecture wth ts applcatons and a threedmenson sensng model are ntroduced n Secton II. Furthermore, n Secton III, a coverage optmzaton algorthm- AFSOA s proposed to deploy sensor nodes n the D envronment. Secton IV presents the smulaton results that the optmal coverage n D WSN s acheved by usng AFSOA, and the algorthm ncreases the coverage rato. Fnally, our conclusons are offered n Secton V /4/$ IEEE 265

2 204 IEEE 6th Internatonal Conference on e-health Networkng, Applcatons and Servces (Healthcom) 2. Archtecture Overvew Hosptal Remote Base Staton BS GPRS/G /4G/5G Wreless Sensor Network II. SYSTEM MODELS Internet Ad Hoc Bluetooth Zgbee Wearable Sensor system WF Dsaster relef center Ter Local Base Staton Ter 2 Ter Fg.. Healthcare montorng archtecture n wreless sensor networks. In Fg., healthcare sensor systems are requred to be connected drectly or ndrectly to the Internet at all tmes, whch allow medcal staffs to tmely acqure arrhythma events and abnormal sgnals for correctng medcal procedures [5]. There are four ters n the proposed healthcare archtecture, ntroduced as follows. Ter WSNs and wearable sensor systems (WSS): Sensor nodes placed n the D survellance regon are selforganzed nto WSNs to gan the envronmental parameters transmtted through a wred or wreless network, communcatng usng Zgbee wreless technology. To capture the ndvduals vtal sgnals, WSS wth Bluetooth wreless transmsson technology are ntegrated wth bomedcal sensor nodes nstalled n patent dentfcaton wrstband or fabrc belt. Ter2 Ad Hoc Network: In ths ter, several wreless routers (WTs) and moble termnals (MTs) are organzed regonally nto an ad hoc network, whch connect to a fxed remote base staton (BS) or local BS through multple hops or an nfrastructure-based network. If BS s destroyed by dsasters, the communcaton command vehcle wll be a substtute for the BS. One MT wth enough computaton capabltes must capture and analyze physcal records from the WSS or WSN because the devce doesn t possess mass data storage capablty over a long perod of tme such as a few months or years. Ter Back-end Network: Ths ter has fxed statons and serves to provde applcaton-level servces for the low ters and process varous sensng data from numerous MTs, whch s structured on the Internet. The server-sde database stores physcalty records from montored ndvduals and ther resdental envronment data for long-term perods. Ter4 Command Network: The data are delvered to the hosptals or dsaster relef center. Then, physcans can provde accurate dagnoses and correct treatment, and commanders gve tmely and accurate orders to the emergency rescue. In each ter, physologcal records are collected or transmtted securely by all knds of protocols or algorthms [6] so that physcans or commanders can provde accurate dagnoses, correct treatment or tmely orders. 2.2 In nursng-house, n-home applcatons The proposed archtecture can be mplemented n a lot of scenaros. In ths secton, we llustrate two mplemented nursng-house and n-home healthcare scenaros usng ths archtecture. Accordng to the nursng-house example n Fg. 2 (left), a lot of sensor nodes deployed n a nursng house can form a wred or wreless network. Based on the lmted power of sensor nodes, some of them should have a sleep mode to stop automatc montorng and reportng n order to reduce power consumpton. Sensor nodes also can organze an alarm network to forward emergency data over all networks when there s a severe envronmental condton. Wearable sensor nodes desgned wth the Zgbee technology could drectly connect to WSNs. Fg. 2 (rght) shows the example of healthcare montorng applcatons at home. It s tmely for wearable sensor nodes on a montored ndvdual to obtan physology sgnals and transmt them tmely to the computng devces whch are held by nurses or famly members. The WSS montor the sustaned physcal postures of the elderly or chronc patents on the computng devces whch perform a healthcare analyss process to search for abnormal fndngs. The applcaton to analyze physcal records wll be appled n the back-end networks for the analyss qualty mprovement. Door Read Back-end Network Read Wred/wreless connecton Staton Wreless sensor node Wearable sensor Fg. 2. Healthcare montorng applcatons n nursng-house and n-home. 266

3 204 IEEE 6th Internatonal Conference on e-health Networkng, Applcatons and Servces (Healthcom) From the analyss of the proposed healthcare montorng archtecture and ts applcatons, we know that how the sensor nodes to deploy n the D space s an urgent ssue to be solved. In the followng secton, we nvestgate a D sensng model. 2. D sensng model In ths paper, a random deployment s consdered wthn a large-scale D WSN. A sphercal omndrectonal coverage model s adopted, and sensor nodes can move n the network. The sensng area of a sensor node s the sphere centered at the sensor node and wth radus R. So, the sensng area of each 4 R sensor node s. We call the pont beng sensed when t locates n the sensng area of sensor nodes. On the study of the contnuous regonal coverage problems n D network, we frst select dscrete ponts wth equal nterval n the coordnate axs X, Y, Z, whch smplfes the problem nto the coverage of dscrete ponts. The set of dscrete ponts s assumed as, and C s the set of dscrete ponts covered by sensor nodes. Then we have C, () C n where (,2,..., n ) s the set of dscrete ponts covered by C sensor node. Then, the coverage rato of the D montorng area s defned as C. (2) We assume the volume and acheved coverage rato of the montorng area are V and a, respectvely. The of sensor nodes dstrbuted n montorng area s N, and the sensng area of those nodes s assumed to be non-overlappng. The relatonshp between the acheved coverage rato and the needed sensor nodes m can be derved from 4 R m a ( ), V () ln( a ) m. ln( 4 R / ( V )) (4) We choose the set wth equal nterval n the X, Y, Z axs drecton. The collecton of n coverage sets s F {,...,,..., } (,2,..., n ), and the weght s C C Cn w ( C ). Then, w( F) n. The goal of ths paper s to fnd a subset C of F to mnmze the sum of weght, whch covers by coverage rato. Namely, t satsfes the followng formulas at the same tme. C /, (5) C C C mn w( ). (6) C C C After the study of heurstc search algorthms-ga, PSOA, ACA and AFSA, we know that the drawbacks of GA, PSOA and ACA lmt ther applcatons n the network coverage optmzaton. In some cases, they even lead to a low coverage rato, the neffectve use of energy and a short lfetme. However, AFSA overcomes the defects of the local mnma of the other algorthms. AFSA s a new knd of ntellgent bonc random searchng algorthm whch s rased to smulate the fsh feedng and survval actvtes to search. It sn t senstve to the selecton of ntal value and parameters, and t s smple, robust, and easy to mplement. At present, AFSA has a benefcal effect on the engneerng feld for a long tme [4]. Therefore, based on ths algorthm, we put forward an artfcal fsh swarm optmzaton algorthm (AFSOA) to fnd out the soluton of (5)-(6), as shown n the followng sectons. III. COVERAGE OPTIMIZATION ALGORITHM In ths secton, we ntroduce AFSOA that adusts the locatons of sensor nodes by AFSA to reduce the overlappng coverage area and make nodes unformly dstrbuted n the D montorng area. Although the coverage rato of the montorng area ncreases, the coverage area s stll redundant. Namely, there are a few redundant nodes n the montorng area. So we pck up the redundant nodes wth the set coverage algorthm, and move them to the uncovered area to mprove the coverage rato. We assume the ntal of sensor nodes s n, and the ntal locaton of sensor node S s x, y, z. So the set of the ntal sensor nodes s I ( x, y, z, x, y, z,..., x, y, z ). The n n n probablty of S to search the target T( xt, yt, z t ) s, d( S, T) Rs Re 2 ( 2 2 p( S, ) ) T e, Rs Re d( S, T) Rs Re, (7) 0, otherwse where R s s the sensng range of sensor nodes, and R c s communcaton radus between sensor nodes., 2,, 2 are the measurement parameters related to the characterstcs of sensor nodes. R (0 R R ) s the measurement e e s relablty parameter. d( S, T ) s the Eucldean dstance of node S and the target T. Re Rs d( S, T), (8) 2 Rs Re d( S, T). (9) We assume S s the set of sensor nodes montorng the target T, and the probablty of target T beng sensed by sensor nodes s p( S, T) ( p( S, T)). (0) n If p th s the probablty threshold of the montored target, the condton of the target beng montored effectvely s mn { p( S, T)} p. () ( xt, yt, zt ) The searchng space of the artfcal fsh swarm s D. In ths paper, we should get the soluton * ( x, y, z, x, y, z,..., x, y, z ) whch makes all n n n ( x, y, z, x2, y2, z2,..., xn, yn, zn ) satsfy the nequalty th 267

4 204 IEEE 6th Internatonal Conference on e-health Networkng, Applcatons and Servces (Healthcom) ( x, y, z, x2, y2, z2,..., xn, yn, zn ). (2) * ( x, y, z, x2, y2, z2,..., xn, yn, zn ) The procedure of AFSOA s as follows:. Search based on AFSA Intally, sensor nodes are randomly deployed n D montorng area. The locaton adustment of sensor nodes s lmted by the neghbor nodes locatons. Usng the network coverage rato as the optmal obectve functon, we search the global optmal dstrbuton through AFSA. In ths secton, we expatate ths algorthm whch makes the sensor nodes deploy unformly n the montorng area. AFSA s an optmzaton algorthm of smulatng the fsh swarm behavor and usng the foragng, clusterng and tralng behavors to search the global optmal soluton quckly. The smulaton results demonstrate that the optmal coverage n D WSN usng AFSA s acheved n a short tme [4]. The food concentraton s the target. Therefore, AFSA has a robust adaptve capacty. () The Foragng Behavor We denote STEP as the maxmum movng step length, and SR s the sensng range of the artfcal fsh swarm. Current status of artfcal fsh swarm s X, and we randomly choose a state X n the vsble doman. If the food concentraton of s greater than the current concentratons, the fsh step forward n ths drecton. Otherwse, we should randomly select a state agan. After several tmes, f t stll can t meet the above condtons, the fsh randomly step forward. Namely, x ( k) x ( k) ( step)( x x ), FC FC x x, () x ( k) x ( k) ( step), FC FC where,2,..., k, and x, are the food concentraton of x and s the th element of X ( k ). (2) The Clusterng Behavor x s the state vector. FC, X FC x, respectvely. x ( k ) There are n partners n the vsble area of artfcal fsh swarm. The center of the of partners s X C. If there are a lot of food and not too crowded n the center of partners, the fsh step forward n the drecton of the center. Otherwse, perform the foragng behavor. () The talgate behavor X max means the status that the maxmum food densty s n the vsble doman of partners. If FCmax FC, the food densty n ths area s the hghest, and t can be set as the center. Otherwse, carry out the foragng behavor. (4) The bulletn board The bulletn board s used to record the optmal ndvdual status of artfcal fsh swarm and the locaton of the food concentraton. After one acton, the artfcal fsh swarm wll be compared wth the bulletn board. If t s superor to the bulletn boards, t wll take the place of the bulletn board. (5) The behavor choce Accordng to the dfferent problems, the artfcal fsh swarm wll evaluate the surrounded envronment and choose the approprate behavor..2 The selecton of redundant nodes After AFSA, t really makes the dstrbuton of sensor nodes unform and the coverage rato mproved. But n fact, there are stll overlaps among the sensng area of those nodes n the D WSN. In other words, there are always a lot of redundant nodes whose selecton s based on the set coverage model that s a NP-hard problem []. In ths paper, to get a decent performance approxmaton algorthm of set coverage problem, we adopt the dea of greedy algorthm to solve t. The complementary set C ' of C conssts of redundant nodes. The desgn of ths algorthm s as follows: select a pont set F every tme, and ts weght s w. To make C w C mnmum, for w, so C s maxmum. We select sensor nodes untl the coverage rato of a pont set s. The algorthm s descrbed as follows: Algorthm : The redundant nodes selecton () The ntal of sensor nodes s 0. (2) Select the maxmum value from {,...,,..., } as C C Cn the frst one. () m, and add the of ths node n the array a. (4) Remove the covered pont from, and get a new pont set. (5) Select the maxmum value of C ' from the remanng nodes as the next one. (6) m, and add the of ths node n the array a. (7) Repeat (4)-(6) untl the covered pont sets are. (8) Return m and the array a.. The movement of redundant nodes In WSNs, to obtan a hgher coverage rato and reduce the blnd spot, we need to ncrease the deployment densty of sensor nodes. However, f the deployment densty s larger, there are more redundant nodes whch cause a large of data transfer conflcts and energy consumpton. Then, the lfetme of WSN s very short. For one sensor node, f ts sensng area can be fully covered by other nodes, we call t redundant. Dfferent from the algorthms [8] that make the redundant nodes enter a dormant state, we move them nto the uncovered area to mprove the whole coverage rato of the montorng area n ths paper. We assume that the of needed nodes s m n practce. Then, the of redundant nodes s N-m. The redundant nodes are moved to the selected uncovered network 268

5 204 IEEE 6th Internatonal Conference on e-health Networkng, Applcatons and Servces (Healthcom) grds. The ultmate target locatons of redundant nodes are the grds whose sensng area contan the hghest of neghbor grds. We elmnate the covered grds n step and repeat ths procedure untl redundant nodes are completely placed. IV.SIMULATION RESULTS In ths secton, a seres of smulaton experments wth Matlab 7.0 are presented to verfy the valdty of the proposed algorthm. We assume that the sze of the D montorng area s m, and there are 85 sensor nodes n the area randomly. The sensng radus of sensor nodes s R 20m, and the sde length of dvded grd s 4m. The comparson between the ntal coverage rato and the optmal coverage rato of the D network s llustrated n Fg.. We can note that the dstrbuton of sensor nodes s more unform after AFSA. The coverage rato ncreases from the orgnal 8.67% to 92.24%. Further, we elmnate the redundant nodes from the optmal network. As a result, the of sensor nodes s from ntal 85 to 75 after the redundancy elmnaton. So the of the redundant nodes s 0. For 0 redundant nodes, we can make them ether sleep or move them nto the uncovered area. In order to mprove the coverage, we adopt to move 0 redundant nodes nto the uncovered area, and the coverage rato ncreases to 96.98%, as shown n Fg.. The coverage rato The ntal coverage AFSA algorthm The redundancy elmnaton The nodes movement Fg.. The comparson of dfferent condtons To acheve a specfed coverage rato, we need to know the condton of redundancy elmnaton. The experments are fnshed wth the ntal 65, 75, 85, 95, 05, respectvely. We record the needed of sensor nodes for the coverage ratos of 70%~ 7%, as shown n Table. To acheve the gven coverage range, the of sensor nodes s 5 n theory, whle the s 6 expermentally. Through the analyss, we get the causes that the theoretcal value s derved n the contnuous space, whle the expermental value s gotten by the dscrete grd coverage. Though there are some devatons, but they are bascally consstent wth the theoretcal values. The ntal Table. The condtons of the nodes redundancy The coverage rato (%) The practcal The redundant To further enhance the coverage rato, we move the redundant nodes to the uncovered area. Table 2 shows that the coverage rato rses after AFSOA, and the of redundant nodes are 5, 9, 0,, 8, respectvely. We also do smlar experments wth R 20m and =7.26%. Then, sensor nodes are needed n order to acheve the rato,.e., the of redundant nodes ncreases. The ntal Table 2. The condtons of the coverage rato The ntal coverage rato(%) The optmal coverage rato(%) The fnal coverage rato (%) To llustrate the advantages of AFSOA for the coverage optmzaton problem n D WSNs, we compare the random dstrbuton algorthm (RDA), GA, PSOA, ACA, AFSOA wth the same smulaton envronment, and the contrast results are shown n Fg.4. We know that the coverage rato of RDA s less than 80% whch can t meet the practcal demand. However, the coverage ratos of GA, PSOA, ACA and AFSOA are more than 90%, the degree of redundancy s small and the network costs are reduced. Among all the algorthms, AFSOA proposed n ths paper s more effcent to realze the global optmzaton deployment, and mantans a superor network coverage rato and balanced energy for a long tme. Thus, t prolongs the network lfetme, and has a wder applcaton scope. Fg. 4. The comparson of dfferent algorthms 269

6 204 IEEE 6th Internatonal Conference on e-health Networkng, Applcatons and Servces (Healthcom) V.CONCLUSION Coverage s one basc topc n D WSNs, whch drectly determnes the QoS of network. In ths paper, we study the random coverage and redundancy problem n the D healthcare system. A healthcare montorng archtecture coupled wth a WSN and WSS s frst presented to montor chronc patents n nursng-house or elderly n ther home. Then, we propose AFSOA to optmze the nodes deployment, whch mproves the network coverage rato. Extensve smulaton results evaluate the effectveness of our proposed algorthm. In ths paper, we take the D coverage nto consderaton, but not the complex envronment and energy balance that play an mportant role n the healthcare system. So our next research feld s the convoluted surface coverage n a D envronment wth the consderaton of the energy balance n WSN. REFERENCES [] Akyldz, I.F., Su, W., Sankarasubramanam, Y., and Cayrc, E.: A survey on sensor networks, Communcatons magazne, IEEE, 2002, 40, (8), pp [2] Yang, G., and Qao, D.: Crtcal condtons for connected-k-coverage n sensor networks, Communcatons Letters, IEEE, 2008, 2, (9), pp [] Chang, C.-Y., Chang, C.-T., Chen, Y.-C., and Chang, H.-R.: Obstacleresstant deployment algorthms for wreless sensor networks, Vehcular Technology, IEEE Transactons on, 2009, 58, (6), pp [4] Zhao, M.-C., Le, J., Wu, M.-Y., Lu, Y., and Shu, W.: Surface coverage n wreless sensor networks, n Edtor (Ed.)^(Eds.): Book Surface coverage n wreless sensor networks (IEEE, 2009, edn.), pp [5] Oktug, S., Khallov, A., and Tezcan, H.: The effects of terran types on D coverage under heterogeneous deployment strateges n Wreless Sensor Networks, n Edtor (Ed.)^(Eds.): Book The effects of terran types on D coverage under heterogeneous deployment strateges n Wreless Sensor Networks (IEEE, 2008, edn.), pp. -6. [6] Tezcan, H., Cayrc, E., and Coskun, V.: A dstrbuted scheme for D space coverage n tactcal underwater sensor networks, n Edtor (Ed.)^(Eds.): Book A dstrbuted scheme for D space coverage n tactcal underwater sensor networks (IEEE, 2004, edn.), pp [7] Andersen, T., and Trthapura, S.: Wreless sensor deployment for D coverage wth constrants, n Edtor (Ed.)^(Eds.): Book Wreless sensor deployment for D coverage wth constrants (IEEE, 2009, edn.), pp. -4. [8] Oktug, S., Khallov, A., and Tezcan, H.: D coverage analyss under heterogeneous deployment strateges n wreless sensor networks, n Edtor (Ed.)^(Eds.): Book D coverage analyss under heterogeneous deployment strateges n wreless sensor networks (IEEE, 2008, edn.), pp [9] Dng, L., and Guan, Z.-H.: Modelng wreless sensor networks usng random graph theory, Physca A: Statstcal Mechancs and ts Applcatons, 2008, 87, (2), pp [0] Mao, Y., Lu, M., Chen, L., Chen, D., and Xe, L.: Dstrbuted energyeffcent locaton-ndependent coverage protocol n wreless sensor networks, Jsuan Yanu yu Fazhan(Computer Research and Development), 2006, 4, (2), pp [] Bhondekar, A.P., Vg, R., Sngla, M.L., Ghanshyam, C., and Kapur, P.: Genetc algorthm based node placement methodology for wreless sensor networks, n Edtor (Ed.)^(Eds.): Book Genetc algorthm based node placement methodology for wreless sensor networks (Cteseer, 2009, edn.), pp [2] Azz, N., Mohemmed, A.W., and Alas, M.Y.: A wreless sensor network coverage optmzaton algorthm based on partcle swarm optmzaton and Vorono dagram, n Edtor (Ed.)^(Eds.): Book A wreless sensor network coverage optmzaton algorthm based on partcle swarm optmzaton and Vorono dagram (IEEE, 2009, edn.), pp [] Gufeng, W., Yong, W., and Xaolng, T.: An ant colony clusterng routng algorthm for wreless sensor networks, n Edtor (Ed.)^(Eds.): Book An ant colony clusterng routng algorthm for wreless sensor networks (IEEE, 2009, edn.), pp [4] Wang Yyue; Lao Hongme; Hu Hengyang, "Wreless Sensor Network Deployment Usng an Optmzed Artfcal Fsh Swarm Algorthm," Computer Scence and Electroncs Engneerng (ICCSEE), 202 Internatonal Conference on, vol.2, no., pp.90,94, 2-25 March 202. [5] Alemdar, H., and Ersoy, C.: Wreless sensor networks for healthcare: A survey, Computer Networks, 200, 54, (5), pp [6] Al Ameen, M., Lu, J., and Kwak, K.: Securty and prvacy ssues n wreless sensor networks for healthcare applcatons, J. Med. Syst., 202, 6, (), pp [7] Xao, F., WANG, R.-c., SUN, L.-., and Wu, S.: Research on the threedmensonal percepton model and coverage-enhancng algorthm for wreless multmeda sensor networks, The Journal of Chna Unverstes of Posts and Telecommuncatons, 200, 7, pp [8] Le, R., Wenyu, L., and Peng, G.: A coverage algorthm for threedmensonal large-scale sensor network, n Edtor (Ed.)^(Eds.): Book A coverage algorthm for three-dmensonal large-scale sensor network (IEEE, 2007, edn.), pp