TELK OMNIKA Indonesian Journal of Electrical Engineering V ol. 16, No . 1, October 2015, pp . 191 199 DOI: 10.11591/telk omnika.v16.i1.8751 191 Optimal Placement of Sensor s in Mission-specific Mobile Sensor Netw orks Hic ham Ouc hitac hen 1,* , Abdellatif Hair 2 , and Najlae Idrissi 3 Sultan Moula y Slimane Univ ersity , F aculty of Sciences and T echniques 1,2 Applied Mathematics and Scientific Calculus Labor ator y 3 Inf or mation Processing Decision Suppor t Labor ator y P .B . 523, Beni-Mellal, Morocco . * h.ouchitachen@gmail.com Abstract Placement of nodes has been the major challenge in wireless sensor netw or ks . The data repor ted from a sensor is only useful when the position of that sensor is f ound. In this conte xt, se v er al techniques ha v e been proposed to conser v e po w er consumption and to prolong the lif etime of the wire less sensor netw or ks . In this paper , w e propose a ne w algor ithm that accur ately finds the best locations of sensors while minimizing the a v er age energy consumed in the netw or k. More precisely , w e consider a cr itical netw or k in each sensor satisfying its o wn missions and depending on its locations . In addition to fulfill their mission, the sensor tr ies to maintain a good neighbor ing nodes quality . w e deter mine the location of node b y using tw o cr iter ia: the cost and the quality of comm unication. The ai m of this w or k is to de v elop a ne w algor ithm so as to solv e the complicated optimization pro b lem posed in this case while minimiz e the total energy consumption. Our sim ulation results demonstr ate that our algor ithm is v er y adv antageous in ter ms of con v ergence to the appropr iate locations . K e yw or ds: energy , node placement, sensor netw or ks , comm unication quality , genetic algor ithms Cop yright c 2015 Institute of Ad v anced Engineering and Science . All rights reser v ed. 1. Intr oduction Wireless Sensor Netw or ks (WSNs) [1, 2] are f orecast to become highly integ r ated to our daily activities [3]. The WSNs are becoming increasingly popular f or mo nitor ing spatial phenom- ena. Indeed, the y are deplo y ed to collect data from the en vironment, process sensed data and tak e action accordingly . T ypical applications of the WSNs include en vironmental control such as fire fighting or mar ine g round erosion, b ut also sensors installation on br idges or b uildings to mon- itor ear thquak e vibr ation patter ns and v ar ious sur v eillance tasks such as intr uder sur v eillance on premises . Y et, it is still v er y ear ly in the lif etime of such systems and man y research challenges e xist, f or instance their computational capabilities , limited stor age , and especially their depen- dence on a limited po w er supplied b y a batter y . The lo w po w er consumption is one of the major challenges that f ace WSNs toda y . Nodes of both sensor netw or ks and con v entional netw or ks such as wireless LANs and cellular netw or ks , f ocus on p ro vid ing a good comm unication quality , or gett ing letter quality ser- vices , e xcept that the nodes of the sensor netw or ks are , moreo v er , interested in the satisf action of their missions . Thus w e call those kinds of netw or ks , cr itical netw or ks , where the quality and satisf action of the mission of each node is v er y impor tant to impro v e the o v er all perf or mance of cr itical netw or ks mission. Monitor ing in mission-cr itical en vironment with the deplo yment of WSNs is one of the most widely-used application areas . Mission-cr itical en vironment implies monitor ing tasks that happened in locations which are difficult to get access to or easy to lead destro y of the netw or k deplo yment: scenar ios in hazard and emergency monitor ing situations such as chemical pollution, en vironment-or iented pollutions and land secur ity . In the near future it can be e xpected that sur v eillance areas will be equ ipped with a r ange of smar t sensors to pro vide par ts of o v er all e v ent detection and ser vice . Hence , in this class of netw or ks , it is impor tant to control the mobility of a node b y jointly consider ing it s mission and comm unication quality , so as to imp ro v e the o v er all perf or mance , b y controlling the mobility prob lem which is m uch difficult to study currently . Receiv ed J uly 2, 2015; Re vised A ugust 18, 2015; Accepted September 2, 2015 Evaluation Warning : The document was created with Spire.PDF for Python.
192 ISSN: 2302-4046 Researchers ha v e conducted in v estigations on WSNs at almost e v er y la y er of the com- m unication protocol stac k. There e xist se v er al papers on efficient routing algor ithms and data agg regation techniques [4, 5], localization techniques [6, 7] and medium access control (MA C) methods [8, 9]. In addition, researchers tak e adv antage of the fle xibility of WSNs b y design- ing protocols that positiv ely aff ect energy consumption. The major ity of researcher papers that appeared until no w consider sensor netw or k to be static. Recently , mobility control or node placement prob lems become the aim of researches and studies in mobile sensor a nd ad hoc netw or ks . In [10] and [11], the modeling tr ansmission po w er consumption w as considered as a tool of studying prob lems of rela y node placements related to minimizing po w er consumption or maximizing netw or k lif etime , via a function of the distance betw een tw o nodes . In [12] and [13], a study of prob lems that place minim um n umber of rela y nodes as long as the constr aints on node connectivity or tr affic demands are satisfied. In [14], a mobility control prob lem in ad hoc netw or ks is studied to maximiz e the throughput based on the IEEE 802.11 throughput analysis in [15] without consider ing the v ar iation of the link capacity according the distance change betw een tw o end nodes . In [16, 18], prob lems that place rela y nodes consider ing the throughput of the netw or k are stu died. In [16], consider ing the probabilistic model f or the positions of nodes , an algor ithm f or the rela y node placement that maximiz es the throughput of the n etw or k is proposed. In [17], joint rela y node placement and assignment prob le m is studie d. Once the rela y node assignment is deter mined, the prob lem in [17] is reduced to the prob lem of finding the optimal position of a single rela y node while other nodes that comm unicate with the rela y node are assumed to be fix ed. In [18], a cascaded netw or k in which rela y nodes f or m a single chained comm unication flo w is considered and a g r adient- based algor ithm is proposed to control the position of rela y nodes to maximiz e the minim um throughput among those of all the links in the chain. The w or k that w e consider in this paper is clear ly diff erent in se v er al cr itical aspects . First, w e consider a netw or k with m ultiple nodes with controllab le mobility in a mesh topology in a tw o-dimensional space . In f act, the e xtension from a single node with controllab le mobility [16, 17, 19, 20] or from the chained topology [18, 21] to m ultiple nodes with controllab le mobility in a mesh topology is not str aightf orw ard. The rest of the paper is organiz ed as f ollo ws . Section 2 deals with mathematical modeling of the prob lem. Section 3 descr ibes n umer ical results . Conlusion is presented in section 4. 2. Resear c h Method 2.1. Model netw ork In this section, w e present our approache to solv e the prob lem of optimal placement of sensors in mob ile wireless sensors netw or k. W e consider a WSNs (Fig.1) containing se v er al sensors managed b y a base station (BS). The nodes satisfy their missions depending on their locations . T o accomplish their mission, the sensor nodes ha v e to maintain a good quality of comm unication with their neighbors , which also depends on their locations . Our objectiv e is to find the best locations of sensor nodes b y de v eloping Sensor’ s Genetic Algor ithm (SGA). The c h oice of Genetic Algor ithms (GA) in this w or k is motiv ated firstly b y finding an algo- r ithm that per mits to solv e the non-con v e x optimization prob lems; GA impose no regular ity on the function studied (contin uity , diff erentiabilit y , con v e xity ...). Secondly b y getting a rob ust algor ithm f or looking f or a solution v er y close to optimal or near-optimalsolution. Thus , in order to solv e the energetic constr aint which is the main challenge of WSNs , the optimization prob lem to be solv ed is f or m ulated as f ollo ws : W e tak e n sensors car acter iz ed each one b y tw o cost funct ions: Mission and comm unication costs . W e then aim to minimiz e the total cost b y choosing the best location of each sensor calculated via SGA. W e suppose that a set of n sensors is deplo y ed in a geog r aphic area of interest to su- per vise a giv en ph ysical phenomenon. The topology of a WSNs is represented b y the g r aph G = ( C ; E ) , where C = f 1 ; 2 ; :::; n g is a set of n sensors and E C C is the set of wireless links betw een the v ar ious sensors . W e denote C v ( i ) the neighbor set of the sensor i . In the belo w , w e present the meanings of the notations used in our modeling. TELK OMNIKA V ol. 16, No . 1, October 2015 : 191 199 Evaluation Warning : The document was created with Spire.PDF for Python.
TELK OMNIKA ISSN: 2302-4046 193 Figure 1. WSN architecture Notations Meaning C set of sensors A set of activ e sensors C v ( i ) neighbor set of sensor i S i area in where each sensor i can mo v e freely ( x i ; y i ) current position of sensor i ( x s i ; y s i ) mission’ s position of sensor i ( x c i ; y c i ) comm unication’ s position of sensor i ( x op i ; y op i ) optimal location of sensor i ( x op bs ; y op bs ) optimal location of base station d ij distance betw een sensors i and j d is distance betw een current and mission’ s position of a sensor i f c ij ( d ij ) cost of comm unication betw een sensors i et j f s i ( d is ) mission cost of sensor i . c comm unication f actor s sur v eillance f actor 2.2. Optimal placement of sensor s T o minimiz e the energy consumed b y all sensors consider ing the joint comm unication and mission costs , w e propose to find the optimal location ( x op i ; y op i ) of each sensor b y solving the f ollo wing optimization prob lem [22]: min f ( x; y ) = i sf s i ( d is ) + i j 2 C v ( i ) cf c ij ( d ij ) (1) Optimal placement of sensors in mission-specific mobile sensor netw or ks (Hicham Ouchitachen) Evaluation Warning : The document was created with Spire.PDF for Python.
194 ISSN: 2302-4046 subject of ( x i ; y i ) 2 S i 8 ( i; j ) 2 C C v ( i ) W e put: f c i ( d ij ) = j 2 C v ( i ) f c ij ( d ij ) (2) V = ( x 1 ; y 1 ; x 2 ; y 2 ; ::::; x n ; y n ) (3) F s ( x; y ) = i f s i ( d is ) (4) F c ( x; y ) = i f c i ( d ij ) (5) F ( V ) = f ( x; y ) (6) So (1) becomes: minF ( V ) = sF s ( V ) + cF c ( V ) (7) subject of V 2 Q n i =1 S i Pr oposed algorithm T o solv e the optimization prob lem giv en b y (7), w e implement the f ollo wing algor ithm: Data: ( x m i ; y m i ) , ( x c i ; y c i ) , i 2 C Result: ( x op i ; y op i ) , i 2 C initialization of population P ; while No con v ergence do P 0 :=Selection of parents in P ; P 0 := Apply the crossing oper ator on P 0 ; P 0 := Apply the m utation oper ator on P 0 ; P 0 := Replace old parents b y their descendants P 0 ; Ev aluate P 0 ; end Algorithm 1: Algor ithm SGA Our algor ithm star ts b y gener ating an initial population P and e v aluating the adaptation of all individuals in initial population. Then the individuals are r andomly selected f or reproduction according to the pr inciple of sur viv al of the fittest. After that the children (or descendants) are gener ated applying the f ollo wing tw o genetic oper ators: crosso v er and m utation. Those children are mo v ed to a ne w population P 0 and will be replaced, in whole or in par t, b y the children of pre vious gener ations . The ne w population of individuals will then tak e o v er from one gener ation to the ne xt, each representing a gener ation iter ation until reaching the stopping cr iter ion. 2.3. Optimal placement of base station 3. Result and Anal ysis In this section, w e pro vide n umer ical results giv en b y our algor ithm SGA used t o find the best locations of sensors . W e set cost functions and par ameters as f ollo ws [22]: f s ij ( d is ) = 5 exp(10 2 d is 1) , f c ij ( d ij ) = 100 exp( 10 12 log 2 (10 6 ( d ij ))) , and C = f 1 ; 2 ; :::; 12 g . W e consider a netw or k in Fig.2, which consists of 12 nodes , where MP= ( x s i ; y s i ) and CP= ( x c i ; y c i ) denotes respectiv ely the mission and comm unication’ s position of each sensor i . TELK OMNIKA V ol. 16, No . 1, October 2015 : 191 199 Evaluation Warning : The document was created with Spire.PDF for Python.
TELK OMNIKA ISSN: 2302-4046 195 Figure 2. Sim ulation topology Figure 3 and Figure 4 illustr ate that as the sur v eillance f actor becomes larger , as each node gets closer to its target location in order to reduce the cost of its mission, as the comm uni- cation quality cost increases . On the other hand, as the comm unication f actor becomes larger , as each node gets closer to each other to reduce the comm unication cost, proceeding the increased mission cost (sur v eillance). Figure 3. V ar iation of comm unication cost Figure 4. V ar iation of mission cost Optimal placement of sensors in mission-specific mobile sensor netw or ks (Hicham Ouchitachen) Evaluation Warning : The document was created with Spire.PDF for Python.
196 ISSN: 2302-4046 Figure 5. P erf or mances of our algor ithm Figure 5 illustr ate the compar a ison of tw o cases with our proposed algor ithm which are as f ollo ws: First case: each node updates its location consider ing only its mission(sur v eillance). Second case: each node updates its location consider ing only its comm unication quality . The best locatio ns of sensors and the base station calculated respectiv ely b y using tw o algor ithms SGA and BGA are represented b y the Figure 6. Figure 6. Best locations of sensors giv en b y SGA Figure 7 sho ws clear ly that, compar ing to the Sim ulated Annealing (SA), our node place- ment algor ithm SGA can pro vide the appropr iate location of each node according to the v ar iation of w eight f actors while minimizing the total w eighted cost f or missions and comm unication qual- ities of all nodes . Our algor ithm SGA is v er y adv antageous in ter ms of con v ergence , this f act is sho wn clear ly in Figure 8. TELK OMNIKA V ol. 16, No . 1, October 2015 : 191 199 Evaluation Warning : The document was created with Spire.PDF for Python.
TELK OMNIKA ISSN: 2302-4046 197 Figure 7. Con v ergence of objectiv e function giv en b y SGA Figure 8. Number of iter ation as function of mission f actor giv en b y SGA 4. Conc lusion In this paper , w e proposed a no v el algor ithm to solv e the optimization prob lem f or m ulated in order to find the optimal locations of sensors in mobile wireless sensor netw or ks , where each node tr ies to minimiz e the w eighted sum of mission and comm unication cost in a distr ib uted w a y . Our approach is based on genetic algor ithms . W e sho w that, compar ing to other techniques , our algor ithm is v er y adv antageous in ter ms of con v ergence to the optimal solution. Numer ical results pro v e that our algor ithm pro vides the appropr iate location of each node b y consider ing jointly the cost of mission and the quality of comm unication. Ref erences [1] I. Akyildiz, T . Melodia, and K. Cho wdhur y , ”A sur v e y on wireless m ultimedia sensor netw or ks , Computer netw or ks , v ol. 51, no . 4, pp . 921-960, 2007. [2] I. Almalka wi, M. Guerrero Zapata, J . Al-Kar aki, and J . Mor illo-P oz o , ”Wireless m ultimedia sensor netw or ks: Current trends and future directions , Sensors , v ol. 10, no . 7, pp . 6662-6717, 2010. [3] K. F o wler , ”The future of sensors and sensor netw or ks sur v e y results projecting the ne xt 5 y ears , in Proc. Sensors Applications Symposium , 2009 (SAS 2009), Ne w Or leans , LA, USA, F eb 2009, pp . 1-6. [4] J .N. Al-Kar aki, R. Ul-Mustaf a, A.E. Kamal, ”Data agg regation and routing in Wireless Sensor Netw or ks: Optimal and heur istic algor ithms , Computer Netw or ks 53 , 2009, pp . 945-960. Optimal placement of sensors in mission-specific mobile sensor netw or ks (Hicham Ouchitachen) Evaluation Warning : The document was created with Spire.PDF for Python.
198 ISSN: 2302-4046 [5] S . Zar ifzadeh, A. Na yy er i, N. Y azdani, A. Khonsar i, H.H. Bazzaz, ”Joint r ange assignment and routing to conser v e energy in wireless ad hoc netw or ks , Computer Netw or ks 53 ,2009, pp . 1812-1829. [6] G. Mao , B . Fidan, B .D .O . Anderson, ”Wireless sensor net w or k localization techniques , Com- puter Netw or ks 51 , 2007, pp . 2529-2553. [7] K. Y eda v alli, B . Kr ishnamachar i, ”Sequence-Based Localization in Wireless Sensor Netw or ks , Mobile Computing, IEEE T r ansactions on 7 (2008) , pp . 81-94. [8] I. Demir k ol, C . Erso y , F . Alagz, ”MA C protocols f or wireless sensor netw or ks: A sur v e y , IEEE Comm unications Magazine 44 , 2006 pp .115- 121. [9] H. Tseng, A. P ang, J . Chen, C . K uo , ”An adaptiv e contention control str ategy f or IEEE 802.15.4-based wireless sensor netw or ks , IEEE T r ansactions on V ehicular T echnology 58 , 2009, pp . 5164-5173 [10] J . P an, Y .T . Hou, L. Cai, Y . Shi, S .X. Shen, ”T opology control f or wireless sensor netw or ks , in A CM Mobicom , 2003, pp . 286-299. [11] E.R. Chittimalla, A. V enkates w ar an, V . Sar angan, R. Achar y a, ”On the use of nodes with controllab le mobility f or conser ving po w er in manets , in IEEE ICDCSW , 2006, pp . 88-93. [12] S . Misr a, S .D . Hong, G. Xue , J . T ang, ”Constr ained rela y node placement in wireless sensor netw or ks to meet connectivity and sur viv ability requirements , in IEEE INFOCOM , 2008, pp . 281-285. [13] B . Lin, P .-H. Ho , L.L. Xie , X. Shen, ”Rela y station placement in IEEE 802.16 j dual-rela y mmr netw or ks , in IEEE ICC , 2008, pp . 3437- 3441. [14] T . Nadeem, S . P ar thasar ath y , ”Mobility control f or throughput maximization in ad hoc net- w or ks , Wireless Comm unications and Mobile Computing 6 (7) , 2006, pp . 951-967. [15] G. Bianchi, ”P erf or mance analysis of the IEEE 802.11 distr ib uted coordinated function, IEEE Jour nal on Selected Areas in Comm unications 18 (3) , 2000, pp . 535-547. [16] A. So , B . Liang, ”Enhancing WLAN capacity b y str ategic placement of tether less rela y points , IEEE T r ansactions on Mobile Computing 6 (5) , 2007, pp . 522-535. [17] A. Sr iniv as , E. Mod iano , ”Joint node placment and assignment f or throughput optimization in mobile bac kbone netw or ks , in IEEE INFOCOM , 2008, pp . 1804-1812. [18] C . Dixon, E.W . F re w , ”Maintaining optimal comm unication chains in robotic sensor netw or ks using mobility control, Mobile Netw or ks and Applicaitons 14 (3) , 2009, pp . 281-291. [19] H.-T . Roh, J .-W . Lee , ”Optimal placement of a rela y node with controllab le mobility in wireless netw or ks consider ing f air ness , in IEEE CCNC , 2010. [20] H.-T . Roh, J .-W . Lee , ”Joint node scheduling and rela y nod e placement in wireless netw or ks with a rela y node with controllab le mobility , Wireless Comm unications and Mobile Computing 12 (8) , 2012, pp . 699-712. [21] H.-T . Roh, J .-W . Lee , ”Comm u nication-a w are position control f or mobile nodes in v ehicular netw or ks , IEEE Jour nal on Selected Areas i n Comm unications 29 (1) , 2011, pp . 173-186. [22] H.-T . Roh, J .-W . Lee , ”Joint Mission and Comm unication A w are Mobility Control in Mobile Ad-Hoc Netw or ks , WiOpt’12: Modeling and Optimization in Mobile , Ad Hoc , and Wireless Netw or ks , Ma y 2012, P aderbor n, Ger man y . pp .124-129. TELK OMNIKA V ol. 16, No . 1, October 2015 : 191 199 Evaluation Warning : The document was created with Spire.PDF for Python.