Movement - Assisted Sensor Deployment

Transcription

1 Intro Self Deploy Vrtual Movement Performance Concluson Movement - Asssted Sensor Deployment G. Wang, G. Cao, T. La Porta Dego Cammarano Laurea Magstrale n Informatca Facoltà d Ingegnera dell Informazone, Informatca e Statstca Sapenza, Unverstà d Roma December 9, 2013 Dego Cammarano Movement - Asssted Sensor Deployment 1 / 48

2 Intro Self Deploy Vrtual Movement Performance Concluson Outlne 1 Introducton 2 Self - Deployment Protocols 3 Vrtual Movement Protocols 4 Performance Evaluatons 5 Consderatons and Open Issues Dego Cammarano Movement - Asssted Sensor Deployment 2 / 48

3 Intro Self Deploy Vrtual Movement Performance Concluson Localzaton Path Plannng Sensng Model Vorono Dagram Localzaton Technques many applcatons such as envronment montorng and target trackng depend on knowng the locaton of sensor nodes each node must determne ts locaton through a locaton dscovery process (ex. GPS) many technques based on receved sgnal strenght, tme dfference of arrval, angle of arrval Dego Cammarano Movement - Asssted Sensor Deployment 3 / 48

4 Intro Self Deploy Vrtual Movement Performance Concluson Localzaton Path Plannng Sensng Model Vorono Dagram Path Plannng problem ntally studed n the robotc area and recently appled to sensor networks combne the known methods to fnd the best moton path and modfed them to explot the dstrbuted nature of sensor networks assume that moble sensors can move wthout any lmtaton usng the exstng technques Dego Cammarano Movement - Asssted Sensor Deployment 4 / 48

5 Intro Self Deploy Vrtual Movement Performance Concluson Localzaton Path Plannng Sensng Model Vorono Dagram Sensng Model each type of sensor has ts unque sensng model characterzed by ts sensng area, resoluton and accuracy sensng area depends on multple factors lke strenght of the sgnals, dstance between the source and sensor, desred confdence level of sensng each sensor node s assocated wth a sensng area represented by a crcle wth the same radus (sotropc sensng models) Dego Cammarano Movement - Asssted Sensor Deployment 5 / 48

6 Intro Self Deploy Vrtual Movement Performance Concluson Localzaton Path Plannng Sensng Model Vorono Dagram Vorono Dagram Vorono Dagram Vorono polygon G 0 of s 0 Dego Cammarano Movement - Asssted Sensor Deployment 6 / 48

7 Intro Self Deploy Vrtual Movement Performance Concluson Localzaton Path Plannng Sensng Model Vorono Dagram Sensng and Communcaton Range sensors can exchange the locaton nformaton by broadcastng some Vorono neghbors of a sensor can be out of ts communcaton range the calculated polygon of ths sensor n not accurate f the sensng range communcaton range the naccurate constructon of Vorono cell wll not affect the detecton of coverage holes f communcaton range sensng range sensors may ms-detect coverage holes Dego Cammarano Movement - Asssted Sensor Deployment 7 / 48

8 Intro Self Deploy Vrtual Movement Performance Concluson VEC VOR Mnmax Termnaton Optmzatons Basc deployment protocols the deployment protocol runs teratvely n rounds n each round sensors broadcast ther locatons and construct ther Vorono polygons f any hole exsts, sensors calculate where to move to elmnate or reduce the sze of the coverage hole Three algorthms are proposed to calculate the target locatons: VEC pushes sensors away from densely covered area VOR pulls sensors to the sparsely covered area Mnmax moves sensors to the center of ther Vorono polygon Dego Cammarano Movement - Asssted Sensor Deployment 8 / 48

9 Intro Self Deploy Vrtual Movement Performance Concluson VEC VOR Mnmax Termnaton Optmzatons VECtor - based Algorthm nspred by the attrbutes of electro-magnetc partcles: when two partcles are too close to each other, an expellng force pushes them apart the vrtual force wll push the sensors away from each other f coverage hole exsts n ether of ther Vorono polygons the fnal overall force on sensors s the vector summaton of vrtual forces from the boundary and all Vorono neghbors VEC s a proactve algorthm that tres to relocate sensors to be evenly dstrbuted Dego Cammarano Movement - Asssted Sensor Deployment 9 / 48

10 Intro Self Deploy Vrtual Movement Performance Concluson VEC VOR Mnmax Termnaton Optmzatons VECtor - Pseudocode (1) Upon enterng Dscovery phase; set tmer to be dscovery nterval; enter Movng phase upon tmeout; broadcast hello after a random tme slot ; Upon enterng Movng phase; set tmer to be movng nterval; enter Dscovery phase upon tmeout; f c = false then call VEC() Done when satsfyng stop crtera; Dego Cammarano Movement - Asssted Sensor Deployment 10 / 48

11 Intro Self Deploy Vrtual Movement Performance Concluson VEC VOR Mnmax Termnaton Optmzatons VECtor - Pseudocode (2) Upon recevng a hello message from sensor s j ; Update N and G ; f G s newly covered then set c = true ; broadcast OK; Only for VEC Upon recevng an OK message from sensor s j ; set c j = true; Dego Cammarano Movement - Asssted Sensor Deployment 11 / 48

12 Intro Self Deploy Vrtual Movement Performance Concluson VEC VOR Mnmax Termnaton Optmzatons VECtor - Pseudocode (3) VEC(); v = 0; for each s j n N do f c true (d ave > d(s, s j )) then F j = (d ave d(s, s j ))/2 ; v = v + F j ; f c = true (d ave > d(s, s j )) then F j = (d ave d(s, s j )) ; v = v + F j ; end f d ave /2 d b (s ) then F b = d ave /2 d b (s ) ; v = v + F b ; do movement adjustment Dego Cammarano Movement - Asssted Sensor Deployment 12 / 48

13 Intro Self Deploy Vrtual Movement Performance Concluson VEC VOR Mnmax Termnaton Optmzatons VECtor - Executon Round 0 Round 1 Round 2 Dego Cammarano Movement - Asssted Sensor Deployment 13 / 48

14 Intro Self Deploy Vrtual Movement Performance Concluson VEC VOR Mnmax Termnaton Optmzatons VORono - Based Algorthm algorthm that pulls sensors to cover ther local maxmum coverage holes f a sensor detects the exstence of coverage holes, t wll move toward ts farthest Vorono vertex (V far ) and stop when V far can be covered Dego Cammarano Movement - Asssted Sensor Deployment 14 / 48

15 Intro Self Deploy Vrtual Movement Performance Concluson VEC VOR Mnmax Termnaton Optmzatons VORono - Example a sensor needs only to check ts own Vorono polygon A s the farthest Vorono vertex of s 0 and d(a, s 0 ) s longer than the sensng range s 0 moves along lne s 0 A to pont B where d(a, B) s equal to the sensng range Dego Cammarano Movement - Asssted Sensor Deployment 15 / 48

16 Intro Self Deploy Vrtual Movement Performance Concluson VEC VOR Mnmax Termnaton Optmzatons VORono - Inaccurate Vorono polygon the maxmum movng dstance s lmted to be at most half of the communcaton range s 0 does not know s 1 so t wll calculate ts local Vorono polygon as the dotted one and vew area around A as a coverage hole t s qute possble t has to move back after t gets to know s 1 Dego Cammarano Movement - Asssted Sensor Deployment 16 / 48

17 Intro Self Deploy Vrtual Movement Performance Concluson VEC VOR Mnmax Termnaton Optmzatons VORono - Consderatons VOR s a greedy algorthm whch tres to fx the largest hole movng oscllatons may occur f new holes are generated due to sensor s leavng to mtgate ths problem s added an oscllaton control that not allow sensors to move backward mmedately before a sensor moves, t frst check whether ts movng drecton s opposte to that n the prevous round Dego Cammarano Movement - Asssted Sensor Deployment 17 / 48

18 Intro Self Deploy Vrtual Movement Performance Concluson VEC VOR Mnmax Termnaton Optmzatons VORono - Pseudocode Notatons: d max : maxmum movng dstance v,f :vector from s to V far VOR(); v = v,f sensng range; shrnk v to be d max f v > d max ; do oscllaton contol; do movement-adjustment Dego Cammarano Movement - Asssted Sensor Deployment 18 / 48

19 Intro Self Deploy Vrtual Movement Performance Concluson VEC VOR Mnmax Termnaton Optmzatons VORono - Executon Coverage durng rounds Dego Cammarano Movement - Asssted Sensor Deployment 19 / 48

20 Intro Self Deploy Vrtual Movement Performance Concluson VEC VOR Mnmax Termnaton Optmzatons Mnmax Algorthm smlar to VOR chooses the target locaton as the pont nsde the Vorono polygon whose dstance to the farthest Vorono vertex s mnmzed (Mnmax pont O m ) based on the dea that a sensor should not be too far away from any of ts Vorono vertces Dego Cammarano Movement - Asssted Sensor Deployment 20 / 48

21 Intro Self Deploy Vrtual Movement Performance Concluson VEC VOR Mnmax Termnaton Optmzatons Algorthm Termnaton - Proof (1) Notatons [x (r) G (r) A (r), y (r) ]: locaton of the sensor s n the r th round : Vorono polygon of s n the r th round, Â(r) : area of the covered part of G (r) Lemma 1 (a) A (r) total = n =1 A r ; (b) A(r+1) total = n  (r) ; =1 Proof. (a) Vorono dagram s a partton of the target feld (b) the summaton of the covered area of the Vorono polygons n the prevous round s also the whole covered area n the current round Dego Cammarano Movement - Asssted Sensor Deployment 21 / 48

22 Intro Self Deploy Vrtual Movement Performance Concluson VEC VOR Mnmax Termnaton Optmzatons Algorthm Termnaton - Proof (2) Lemma 2 A (r) = A (r) ([x (r), y (r) ]) Proof. Ths s the drect result of the attrbute of Vorono dagram. Every pont wthn G (r) s closer to [x (r), y (r) ] than to any other sensor. Any pont not covered by s s also not be covered by any other sensor. Dego Cammarano Movement - Asssted Sensor Deployment 22 / 48

23 Intro Self Deploy Vrtual Movement Performance Concluson VEC VOR Mnmax Termnaton Optmzatons Algorthm Termnaton - Proof (3) Theorem 1 A (r+1) total Proof. > A (r) total before all sensors stop movng. At the (r + 1) th round, there may be areas n G (r) whch s not covered by s, but s covered by other sensors, because the current Vorono polygon of s s G (r+1) but not G (r). Therefore  (r) A (r) ([x (r+1), y (r+1) ]) (1) By enforcng the movement adjustment heurstcs, VEC, VOR and Mnmax algorthms guarantee that, f s moves, A (r) ([x (r+1), y (r+1) ]) > A (r) ([x (r), y (r) ]) (2) Dego Cammarano Movement - Asssted Sensor Deployment 23 / 48

24 Intro Self Deploy Vrtual Movement Performance Concluson VEC VOR Mnmax Termnaton Optmzatons Algorthm Termnaton - Proof (4) Proof (contnue). Certanly, f s 1 does not move, A (r) because [x (r+1) ([x (r+1) From (1),(2),(3) n =1 Â (r), y (r+1) n =1, y (r+1) ]) = A (r) ] = [x (r), y (r) ] A (r) f some sensor moves n the r th round ([x (r), y (r) ]) (3) ([x (r), y (r) ]) (4) Dego Cammarano Movement - Asssted Sensor Deployment 24 / 48

25 Intro Self Deploy Vrtual Movement Performance Concluson VEC VOR Mnmax Termnaton Optmzatons Algorthm Termnaton - Proof (5) Proof (contnue). From Lemma 2 : A (r) From (4),(5): n =1 Â (r) By Lemma 1: From (6),(7): A (r+1) total = A (r) > n =1 > A (r) total ([x (r), y (r) ]) (5) A (r) (6) A (r) total = A (r+1) total = n =1 n =1 A (r) Â (r) (7) Dego Cammarano Movement - Asssted Sensor Deployment 25 / 48

26 Intro Self Deploy Vrtual Movement Performance Concluson VEC VOR Mnmax Termnaton Optmzatons Algorthm Termnaton - Proof (6) Corollary 1 VEC, VOR and Mnmax algorthms are convergent and thereby termnate naturally Proof. Followng from Theorem 1 and the fact that A (r) total total s upper bounded by the total area of the target feld, dstrbuted algorthms converge, and termnate naturally. All sensors stop movng when no coverage ncrease can happen. Dego Cammarano Movement - Asssted Sensor Deployment 26 / 48

27 Intro Self Deploy Vrtual Movement Performance Concluson VEC VOR Mnmax Termnaton Optmzatons Dealng wth message loss Problem Hello messages may be lost due to collsons sensors may fal to know the presence of some neghbors and mstakenly determne coverage holes Soluton assocate each tem n a sensor s neghbor lst wth a number whch ndcates the freshness of ths tem (.e for how many rounds ths neghbor has not been heard) on constructng Vorono polygon every sensor only consder the sensors n ts neghbor lst wth certan freshness Dego Cammarano Movement - Asssted Sensor Deployment 27 / 48

28 Intro Self Deploy Vrtual Movement Performance Concluson VEC VOR Mnmax Termnaton Optmzatons Dealng wth poston clusterng Problem the ntal deployment of sensors may form clusters resultng n a low ntal coverage sensors located nsde the clusters can not move for several rounds the deployment tme s prolonged and some sensors can stll be clustered after several rounds Soluton detecton of stuatons n whch many sensors are clustered n small area durng the frst round the algorthm explodes cluster to scatter the sensor apart each sensor after concludes that t s nsde a cluster choose a random poston wthn an area centered tself Dego Cammarano Movement - Asssted Sensor Deployment 28 / 48

29 Intro Self Deploy Vrtual Movement Performance Concluson Protocol Implementaton Pseudocode A frst approach The basc protocols requre sensors to move teratvely, eventually reachng the fnal destnaton. Other approaches can be envsoned n whch the sensors move only once to ther destnaton to mnmze the sensor movement. sensors calculate ther target locatons, vrtually move there exchange these new vrtual locatons wth the sensors whch would be ther neghbors as f they had actually moved. the real movement only happens at the last round after fnal destnatons are determned Drawbacks susceptble to poor performance under network parttons hgh communcaton overhead Dego Cammarano Movement - Asssted Sensor Deployment 29 / 48

30 Intro Self Deploy Vrtual Movement Performance Concluson Protocol Implementaton Pseudocode A better approach Idea To get balance between movement and message complexty, sensors do vrtual movement when the communcaton cost to reach the logcal Vorono neghbors s reasonable, and do physcal movement otherwse. The challenge s to determne f a sensor can reach ts logcal neghbors wth reasonable communcaton cost. Dego Cammarano Movement - Asssted Sensor Deployment 30 / 48

31 Intro Self Deploy Vrtual Movement Performance Concluson Protocol Implementaton Pseudocode Heurstc 1 f sensor s dstance to ts farthest Vorono vertex s shorter then half of the communcaton range, t must know all ts Vorono neghbors one hop broadcast s enough to exchange the locaton nformaton wth ts logcal neghbors physcal movement s not necessary 2 some Vorono neghbors are out of the communcaton range sensors request ther neghbors lsts to obtan the logcal postons of sensors wthn two broadcast hops when the dstances between the physcal locatons of sensors and ther farthest Vorono vertces are larger than two tmes the maxmum movng dstance, sensors should move physcally Dego Cammarano Movement - Asssted Sensor Deployment 31 / 48

32 Intro Self Deploy Vrtual Movement Performance Concluson Protocol Implementaton Pseudocode Implementaton 1 Dscovery phase sub-phase 1: sensors broadcast hello messages sub-phase 2: sensors broadcast the locaton of known neghbors 2 Logcal movng phase f a sensor s dstance to ts farthest Vorono vertex s larger than half of the communcaton range, calculate the target locaton and do logcal movement set a flag n the hello messages to requre broadcastng of neghbors lst any sensor that receves a hello message wll broadcast ts neghbor lst n the second sub-phase of the dscovery phase 3 Physcal movng phase ts physcal poston s two tmes the maxmum movng dstance to ts farthest Vorono vertex a sensor s logcal poston has not changed for several rounds Dego Cammarano Movement - Asssted Sensor Deployment 32 / 48

33 Intro Self Deploy Vrtual Movement Performance Concluson Protocol Implementaton Pseudocode Vrtual Algorthm - Pseudocode (1) Upon enterng Dscovery phase-i; set tmer to be dscovery nterval/2; enter Dscovery phase-ii ; broadcast hello(w ) after a random tme slot; Upon enterng Dscovery phase-ii; set tmer to be dscovery nterval/2; enter Movng phase upon tmeout; f l = true then broadcast NL after a random tme slot; l = false; Dego Cammarano Movement - Asssted Sensor Deployment 33 / 48

34 Intro Self Deploy Vrtual Movement Performance Concluson Protocol Implementaton Pseudocode Vrtual Algorthm - Pseudocode (2) Upon enterng Movng phase; set tmer to be movng nterval; enter Dscovery phase-i upon tmeout; f c = false then calculate loc by VEC or VOR or Mnmax; do oscllaton control; do movement adjustment; p = 1 f logcally moves; f d(loc, V far ) 2 d max then move to loc ; w = false; f d(loc, V far ) d c /2 then move to loc ; w = true;... Dego Cammarano Movement - Asssted Sensor Deployment 34 / 48

35 Intro Self Deploy Vrtual Movement Performance Concluson Protocol Implementaton Pseudocode Vrtual Algorthm - Pseudocode (3) Upon recevng a hello(w j ) message from sensor s j ; f w j = true then l = true; Update N and G ; f G s newly covered then set c = true; Upon recevng a NL j message from sensor s j ; Update N and G ; Dego Cammarano Movement - Asssted Sensor Deployment 35 / 48

36 Intro Self Deploy Vrtual Movement Performance Concluson Objectves Methodology Smulaton Results Objectves testng the effectveness of protocols n provdng hgh coverage comparng VEC, VOR and Mnmax comparng the Basc protocols and the Vrtual movement protocols studyng the effectveness of controllng the tradeoff among varous metrcs by adjustng parameters Dego Cammarano Movement - Asssted Sensor Deployment 36 / 48

37 Intro Self Deploy Vrtual Movement Performance Concluson Objectves Methodology Smulaton Results Metrcs 1 Deployment qualty sensor coverage tme (number of rounds) to reach ths coverage 2 Energy consumpton mechancal movement (startng/brakng energy and movng energy) communcaton Dego Cammarano Movement - Asssted Sensor Deployment 37 / 48

38 Intro Self Deploy Vrtual Movement Performance Concluson Objectves Methodology Smulaton Results Methodology smulatons done under dfferent sensor densty to determne the sensor coverage that can be reached and the dffculty to reach t a 100m*100m target feld, 4 dfferent number of sensors dstrbuted (from 120 to 180) run 10 experments based on dfferent ntal dstrbuton and calculate the average results MAC layer protocol, DSDV routng protocol based on Bellman-Ford algorthm 6 meters sensng range, 20 meters communcaton range Dego Cammarano Movement - Asssted Sensor Deployment 38 / 48

39 Intro Self Deploy Vrtual Movement Performance Concluson Objectves Methodology Smulaton Results Coverage VEC s senstve to ntal deployment VOR and Mnmax both move to heal the holes drectly and acheve qute smlar coverage Basc protocols and Vrtual movement protocols acheve almost the same coverage Dego Cammarano Movement - Asssted Sensor Deployment 39 / 48

40 Intro Self Deploy Vrtual Movement Performance Concluson Objectves Methodology Smulaton Results Energy Consupton - Basc vs Vrtual Protocols Basc Vrtual Dego Cammarano Movement - Asssted Sensor Deployment 40 / 48

41 Intro Self Deploy Vrtual Movement Performance Concluson Objectves Methodology Smulaton Results Message Complexty - Basc vs Vrtual Protocols Message complexty s the number of messages exchanged when the protocol termnates and represents the normalzaton of the movng dstance and the number of movements Basc Vrtual Dego Cammarano Movement - Asssted Sensor Deployment 41 / 48

42 Intro Self Deploy Vrtual Movement Performance Concluson Objectves Methodology Smulaton Results Convergence tme Dego Cammarano Movement - Asssted Sensor Deployment 42 / 48

43 Intro Self Deploy Vrtual Movement Performance Concluson Objectves Methodology Smulaton Results Termnaton Dego Cammarano Movement - Asssted Sensor Deployment 43 / 48

44 Intro Self Deploy Vrtual Movement Performance Concluson Consderatons Open Issues Consderatons VEC move least at low sensor densty and can be deployed when the coverage requrements s not hgh VOR termnates the earlest when the sensor densty s not very hgh and t can be deployed when both the deployment tme requrement and coverage requrement are strct Mnmax s the best choce to calculate the target locaton Vrtual movement protocols can sgnfcantly reduce mechancal movement wth a cost of less then two Dego Cammarano Movement - Asssted Sensor Deployment 44 / 48

45 Intro Self Deploy Vrtual Movement Performance Concluson Consderatons Open Issues Dstrbuted scheme sensors calculate ther target locaton followng a dstrbuted approach sensors only move when there are coverage holes can t always guarantee the optmal coverage Dego Cammarano Movement - Asssted Sensor Deployment 45 / 48

46 Intro Self Deploy Vrtual Movement Performance Concluson Consderatons Open Issues Centralzed scheme a centralzed approach may not be fesble n some deployments and suffers from the problem of a sngle pont of falure optmal poston to place sensors s decded a pror by a central server who mnmze the movng dstance of the sensors ths s a bpartte matchng problem and the classc Hungaran method s used to dstrbute sensors mnmzng ther movng dstance Dego Cammarano Movement - Asssted Sensor Deployment 46 / 48

47 Intro Self Deploy Vrtual Movement Performance Concluson Consderatons Open Issues Sensng Area Area shape the sensng area of each sensor s a dsk wth radus 6m f the sensng area s unformly a dsk, protocols can deal well wth the case of a larger or smaller sensng radus f the sensng area s an rregular shape, sensors can stll determne the coverage holes by reducng the sensng range Performance performance depends more on the rato of communcaton range to sensng range than the absolute sensng range on decreasng of the sensng range, protocols can accurately construct the Vorono dagrams on ncreasng of the sensng range, t s necessary to enlarge broadcast hops to better construct the Vorono polygons Dego Cammarano Movement - Asssted Sensor Deployment 47 / 48

48 Intro Self Deploy Vrtual Movement Performance Concluson Consderatons Open Issues Questons Dego Cammarano Movement - Asssted Sensor Deployment 48 / 48