Centralzed approach for mult-node localzaton and dentfcaton Ola A. Hasan Electrcal Engneerng Department Unversty of Basrah Basrah, Iraq Lolastar91@gmal.com Ramzy S. Al Electrcal Engneerng Department Unversty of Basrah Basrah, Iraq rsawaly@gmal.com Abdulmuttalb T. Rashd Electrcal Engneerng Department Unversty of Basrah Basrah, Iraq abdturky@gmal.com Abstract: A new algorthm for the localzaton and dentfcaton of mult-node systems has been ntroduced n ths paper; ths algorthm s based on the dea of usng a beacon provded wth a dstance sensor and IR sensor to calculate the locaton and to know the dentty of each vsble node durng scannng. Furthermore, the beacon s fxed at mddle of the frame bottom edge for a better vson of nodes. Any detected node wll start to communcate wth the neghborng nodes by usng the IR sensors dstrbuted on ts permeter; that nformaton wll be used later for the localzaton of nvsble nodes. The performance of ths algorthm s shown by the mplementaton of several smulatons. Index Terms Beacon node, Identfcaton, Localzaton, Mult-node. I. INTRODUCTION Wreless sensor networks have attracted a lot of attenton due to ther mportant role and ncreased utlzaton n many mltary and cvlan domans [1-3]. Localzaton s one of the most mportant challenges n the development of wreless sensor networks [4-5], the reason for that s, collectng data wthout locatons wll make these data geographcally meanngless [6-7]. Wreless sensor network consst of many sensor nodes that each contans sensors, battery, processor and other devces that may be needed [8]. Every node n wreless sensor network collects data and ether forwards them to a central node to calculate ther locatons and ths wll be called a centralzed archtecture [9-10] or use these data to localze tself and then forward the data and ths wll be called a dstrbuted archtecture [11-12]. Localzaton can be relatve, absolute or a combnaton of them [13-14].Relatve localzaton s usually based on the dead reckonng (that s, montorng the wheel revolutons to fnd the dstance from a known startng locaton). The unbounded accumulaton of errors s the man dsadvantage of ths method and n the case of movement, relatve localzaton s preferable. On the other hand, absolute localzaton depends on the satellte sgnals, beacons, landmarks or map matchng to fnd the locaton of a node. GPS s one of the used methods as an example of the absolute localzaton but t can be used only outdoor and t s naccurate for moble nodes [15-16]. In ths paper, we ntroduce a combnaton of both absolute and relatve localzaton; the absolute localzaton s represented by the beacon whch scans the envronment lookng for vsble nodes. On the other hand, the relatve localzaton s generated by vsble nodes whle explorng the envronment n search of ther neghbors. Also, n ths paper we use the centralzed archtecture where all the nformaton results from scannng the envronment by both of the beacon and vsble nodes s collected n the beacon to construct two tables whch wll be used later to localze the nvsble nodes. The detals of ths algorthm wll be dscussed n secton II and the related work n secton III; secton IV shows the smulaton results of ths algorthm. Fnally the concluson wll be n secton V. II. RELATED WORK When we talk about obtanng locaton nformaton, the frst thng came nto our mnd s the global postonng system (GPS) proposed by B. Hofmann-Wellenhof [17]. But due to ts hgh cost and n ablty to work n ndoor envronment, a lot of algorthms concerns wth localzaton 178
accordng to the type of nformaton used to localze nodes have been studed [9]. Many of those algorthms lay on the range-based schemes whch use the ranged measurements strateges such as the receved sgnal strength (RSS), angle of arrval (AOA), tme of arrval (TOA) and tme dfference of arrval (TDOA) to compute ether the dstance or angle between two nodes [18]. As an example s the algorthm proposed by P. Bahl et al. [19] whch s a method to convert the RSS to a dstance and then uses the trangulaton to calculate the node's poston. Some other papers lay on the range free strateges such as: approxmate pont n trangle (APIT), DV-Hop and centrod. Those strateges take nto account the connectvty among nodes to acheve localzaton [9]. For example, centrod algorthm proposed by N. Bulusu et al. [20] whch uses a centrod formula to localze the nodes after recevng the locatons of beacons. The range based algorthms are more accurate than the range free but the range free algorthms are less n the term of complcty and cost. Therefore, some researchers focused on proposng methods that combne the benefts of the range based and range free both and consdered to be more accurate than the range free and less cost and complcty than the range based. An example for that s the algorthm proposed by A. T. Rashd et al. [21] where a beacon s provded wth dstance sensor to determne the locatons of vsble nodes and construct clusters; those clusters wll be compared later wth unt dsk graph for each node whch constructed by the connectvty among nodes to know the dentty of them. In ths paper we wll propose an algorthm that s better than the algorthm proposed by A. T. Rashd et al. n the case of vsblty, accuracy and computaton complcty. III. LOCATION AND IDENTITY COMBINATION ALGORITHM In ths algorthm, the beacon whch s located at mddle of the frame bottom edge contans a dstance sensor and IR sensor par (transmtter and recever) as shown n Fg. 1; those sensors are fxed on a servo motor whch rotates at 180⁰. Also, there are n of IR sensors dstrbuted evenly on the permeter of every node n the system. The locaton and dentty of each node are found accordng to the followng subsectons: Fg.1, The constructon of the Beacon node. A. Localzaton of Vsble Nodes: The beacon scans the envronment at 180⁰ searchng for any node as shown n Fg.2, f the beacon detects a node, the coordnates of ths node wll be calculated as n the equaton below: x = x b + d b * cos α b y = y b + d b * sn α b (1) Where d b and α b are the dstance and the detectng angle of node respectvely, (x b, y b ) are the coordnates of the beacon and (x, y ) are the coordnates of node. B. Identty of Vsble Nodes: After detectng a node, the beacon uses the IR sensor par to send an nfrared sgnal to dentfy tself and wats for the node to reply wth ts dentty number. Each node s provded wth n of IR sensor pars whch numbered n sequentally form startng from the reference IR sensor as shown n Fg.3; the angle K between each two neghborng sensor pars are equal and they are computed as n equaton2: K = 360 / n (2) 179
C. Constructon of beacon vsblty based table: After scannng the entre envronment, the beacon starts to construct a table whch contans the dentty, detectng angle and the coordnates of each vsble node. As an example the table s contents for vsble nodes shown n Fg. 2 wll look as shown n table 1. Table 1.The beacon vsblty based table Node ID Detecton angle x y ID8 αb 8 x 8 y 8 ID1 αb 1 x 1 y 1 ID4 αb 4 x 4 y 4 ID7 αb 7 x 7 y 7 ID5 αb 5 x 5 y 5 ID9 αb 9 x 9 y 9 Fg.2, Scannng the envronment wth beacon node. Each node n the envronment sequentally actvates ther IR recever sensor watng for any sgnals come from the beacon. If any node detects a sgnal that represent the ID number of beacon then ths node replays by sendng ts ID number through the transmtter sensor of the same IR sensor par that has receved the beacon sgnal. ID6 αb 6 x 6 y 6 D. Constructon of Neghborng Nodes Table: Durng the constructon of the beacon vsblty based table, there s also another table beng constructed n the beacon whch s called the neghborng nodes table ;ths new table shows the neghborng nodes for each vsble node and t wll be used later to localze the nvsble nodes. The followng are the constructon steps: 1. If the beacon detects an object, t wll send ts ID number (IDBn) and wat wth a specfc tme to receve a reply.otherwse the beacon wll move to the next angle degree. Fg.3. Schematc for node wth 8 IR sensor pars (Transmtter and Recever). 2. If the detected object s a node, then t wll reply wth ts ID number frst and start to scan searchng for neghbors usng ts IR sensor pars. The process of scannng s acheved by sequentally sendng the ID number of the node from each IR sensor par and wats for the ID number of the neghbor node. The IR sensor par s marked by Null value f t does not have any replay. The beacon stays n a wat state untl the recepton of ths node's 180
scannng data. Fg.4 s an example that shows step 1 and a part of step 2. 3. After the completon of step 2 a new raw s added to the neghborng nodes table contans the neghbor nodes to that detected node. 4. The beacon contnues to repeat all the steps above untl ts rotaton reaches 180. 5. When the beacon completes ts scannng the neghborng nodes table for vsble nodes wll be completed as showng n table 2. locatons of those two neghbors and the orentaton of one of them. The process s acheved accordng to the followng steps: 1. The beacon can conclude whch nodes are nvsble n table 1 and from table 2 the beacon can choose two of ther vsble neghbors whch ther coordnates are located n table 1. Fg. 6 shows node 2 whch s nvsble to the beacon and two of ts vsble neghbors whch are node 4 and node 5. Fg.5 shows the connectvty among the nodes n the envronment wtch ther nformaton s mentoned n table 2. The connectvty may be defned as the representaton of nodes n a unt dsk graph [21]. In unt dsk graph we can use the absolute locatons and orentaton of some other nodes connected to the node whch we want to estmate ts locaton and orentaton. The estmaton accuracy ncreases as the connectvty of unt dsk graph ncreases. Fg.5. llustrates the connectvty among nodes n table 2. Fg.4. Schematc for sngle node wth 8 IR Tr. And Rec.. E. Localzaton of Invsble Nodes: The locaton of any nvsble node n the system s found wth the help of nformaton obtaned from two of ts vsble neghbors. Ths nformaton ncludes 2. To compute the locaton of nvsble node, the beacon must at frst compute the orentaton of one of ts two vsble neghbors that have been chosen. The orentaton means the angle between X- axs and the drecton of reference IR sensor par. 3. From table 1 the beacon chooses the detectng angle α b 5 of the vsble node 5. 4. From Fg. 6, the orentaton angle β 5 of node 5 s computed accordng to the followng equatons: ϕ 5 14 = (4-1) * k (3) Where ϕ 5 14 s the angle between IR 5 1 and IR 5 4 sensors on node 5. 181
σ = 180 - ϕ 5 14 (4) The coordnates (x 2, y 2 ) of the nvsble node 2 wll be as below: β 5 = σ + α b 5 (5) 5. The localzaton of nvsble node 2 as n Fg.7 s acheved accordng to set of equatons as follow: x 2 = x 5 + R * cos θ y 2 = y 5 + R * sn θ (13) L = ((y 5 y 4 ) 2 + (x 5 x 4 ) 2 ) 1/2 (6) Where L s the dstance between node 4 and node 5. ϕ 5 56 = (6 5) * K (7) Where ϕ 5 56 s the angle between IR 5 5 (the sensor whch receved the sgnal from node 4) and IR 5 6 (the sensor whch receved the sgnal from node 2) on node 5. ϕ 4 12 = (2 1) * K (8) Where ϕ 4 12 s the angle between IR 4 1 (the sensor whch receved the sgnal from node 2) and IR 4 2 (the sensor whch receved the sgnal from node 5) on node 4. ϕ 2 82 = 180 - (ϕ 5 56 + ϕ 4 12 ) (9) ϕ 2 82 represents the angle between the IR sensors whch receved the sgnals from nodes 5 and 4. By usng the sn low: L / sn ϕ 2 82 = R / sn ϕ 4 12 (10) Where R s the dstance between node 5 and node 2. ϕ 5 61 = (9 6) * K (11) ϕ 5 61 s the angle between the orentaton of node 5 and the dstance between node 2 and node 5. The number 9 represents (8+1), where 8 s total number of IR sensor pars on each node. θ = β 5 ϕ 5 61 (12) Fg.6, the orentaton of node 5. Fg.7. Localzaton of node 2. 182
Table 2. The neghborng nodes table IR 1 IR 2 IR 3 IR 4 IR 5 IR 6 IR 7 IR 8 ID8 Null Null Null ID1 ID4 Null IDBn Null ID1 ID8 Null Null Null Null ID3 ID4 IDBn ID4 ID2 ID5 ID7 IDBn ID8 ID1 Null ID3 ID7 Null ID9 ID6 IDBn Null Null ID4 Null ID5 Null Null ID9 IDBn ID4 ID2 Null Null ID9 ID6 IDBn ID7 ID3 ID5 Null Null Null ID6 Null IDBn ID7 ID9 Null Null Null Null IV. THE SIMULATION RESULTS Numercal smulatons have been mplemented by usng vsual basc 2012 programmng language. It s repeated over 100 tmes on dfferent szes of networks rangng from 5 to 50 nodes and also repeated for three nodes' raduses whch are 10, 15 and 20 pxels. Nodes are dstrbuted on an area of 500*500 pxels and each node has a unque ID number. The results of our algorthm are compared wth the robotc cluster matchng algorthm [21]; the robotc cluster matchng algorthm uses a combnaton of absolute and relatve sources for localzaton and orentaton of mult-robot systems. The absolute nformaton s obtaned from the beacon whch s a dstance sensor located at the left bottom corner of the frame and rotates at 90⁰. The beacon scans the envronment for vsble robots that wll be used to form clusters. On the other hand, the relatve nformaton s obtaned from robots where every robot scans the envronment lookng for ts neghbors to construct a unt dsk graph. Fnally, by the matchng of clusters and the unt dsk graph of each robot the vsble robots wll be localzed. In our algorthm the beacon s provded wth a dstance and IR sensors; t s located at mddle of the frame bottom edge and rotates at 180⁰. Fg.8, Fg. 9, and Fg. 10 study the effects of 10, 15 and 20 pxels node radus respectvely on the vsblty (the number of vsble nodes) of 30 nodes envronment, where the black nodes are vsble to the beacon whle the gray ones are not. It s obvous that the vsblty decreases as the radus of nodes ncreases. The ncreasng n vsblty means that the number of nodes whch localzed by the beacon ncreased and thus leads to an ncrease n number of nodes whch have accurate locatons. Fg. 8, Illustraton the effects of nodes wth 10 pxels radus on 30 nodes envronment Fg.11 shows a comparson between the vsblty percentage of our algorthm and the robotc 183
cluster matchng algorthm wth dfferent number of nodes each of 10 pxels radus. Fg. 11, the vsblty comparson between the locaton and dentty combnaton algorthm and the robotc cluster matchng algorthm wth nodes has 10 pxels radus. Fg. 9, Illustraton the effects of nodes wth 15 pxels radus on 30 nodes envronment Fg.12 and Fg. 13 show the same comparson but wth 15 and 20 pxels nodes radus respectvely. From these fgures we notce that the vsblty percentage of both algorthms decrease as the radus and the number of nodes ncrease but wth all raduses and wth all number of nodes our algorthm shows a better performance than the robotc cluster matchng algorthm. Fg. 10, Illustraton the effects of nodes wth 20 pxels radus on 30 nodes envronment Fg. 12, the vsblty comparson between the locaton and dentty combnaton algorthm and the robotc cluster matchng algorthm wth nodes has 15 pxels radus. 184
Fg. 13, the vsblty comparson between the locaton and dentty combnaton algorthm and the robotc cluster matchng algorthm wth nodes has 20 pxels radus. In Fg. 14, Fg. 15 and Fg. 16 the effects of the beacon rotatng angle and the offset between the beacon and vsble nodes on the localzaton accuracy have been studed. Fg.14 shows an envronment of 30 nodes each of 15 pxels radus nodes and a beacon wth 1 degree rotaton angle. Fg.15 and Fg. 16 show the same envronments but wth 2 and 3 degrees rotaton angle respectvely. Fg.15. Medan error n locaton estmaton for nodes wth 15 pxels radus and beacon wth 2 degree rotaton angle. Fg.16. Medan error n locaton estmaton for nodes wth 15 pxels radus and beacon wth 3 degree rotaton angle. Fg.14. Medan error n locaton estmaton for nodes wth 15 pxels radus and beacon wth 1 degree rotaton angle. Fg.17 shows that by ncreasng the beacon rotaton angle, the error n locaton estmaton wll also ncrease and thus wll reduce the accuracy. It s worth to menton here, that ncreasng the offset between the beacon and 185
vsble nodes wll also cause an ncreasng n the error average. suggeston s to add a beacon at mddle of the frame upper edge to have more vsblty average. REFERENCES Fg. 17, the accuracy of localzaton for dfferent rotatng angles. V. CONCLUSIONS In ths paper, an algorthm for mult-node localzaton system has been ntroduced; t uses the dea of centralzed archtecture where all the locatons computaton s done n a centralzed staton whch s the beacon. In ths algorthm, the beacon also serves as a source of absolute nformaton durng the envronment scannng n search of vsble nodes. The source of relatve nformaton s represented by the nodes where each node scans the envronment to fnd ts neghbors. The poston of beacon n our algorthm at mddle of the frame bottom edge has hghlghted the mportant role of ths poston on the nodes vsblty average. So, as compared wth the robotc cluster matchng algorthm, our algorthm shows an ncrease n the vsblty average meanng that more accuracy n locaton estmaton wll be obtaned. Also, our proposed algorthm shows a better performance than the robotc cluster matchng algorthm n addressng the nodes under the effects of dfferent parameters such as the rotatng angle of beacon, nodes radus and the sze of network. The proposed algorthm n ths paper deals wth nodes only. So, as an mprovement we suggest to add a separate secton for the orentaton calculaton to nvolve even the robots. Another [1] M. L, and Y. Lu, "Angle of arrval estmaton for localzaton and communcaton n wreless networks," 16th European sgnal processng conference (EUSIPCO 2008), pp. 25-29, 2008. [2] Y. Bokareva, W. Hu, S. Kanhere, B. Rstc, N. Gordon, T. Bessell, M. Rutten and S. Jha, "Wreless Sensor Networks for Battlefeld Survellance," Land Warfare Conference 2006. [3] Q. I. Al, A. Abdulmaowjod and H. M. Mohammed, "Smulaton and performance study of wreless sensor network (WSN) usng MATLAB," Iraq J. Electrcal and electroncal engneerng, vol.7, no. 2, pp. 112-119, 2011. [4] N. Dhopre, and Sh. Y. Gakwad, "Localzaton of wreless sensor networks wth rangng qualty n woods," Internatonal journal of nnovatve research n computer and communcaton engneerng, vol. 3, no. 4, pp.2893-2896, 2015. [5] P.K. Sngh, B. Trpath, and N. P. Sngh, "Node localzaton n wreless sensor network," Internatonal Journal of Computer Scence and Informaton Technologes, vol. 2, no. 6, pp. 2568-2572, 2011. [6] G. S. Klogo,and J. D. Gadze, "Energy Constrants of Localzaton Technques n Wreless Sensor Networks (WSN): A Survey," Internatonal Journal of Computer Applcatons, vol. 75, no. 9, pp. 0975 8887, March 2013. [7] J. Zhang, H. L and J. L, "An Improved CPE Localzaton Algorthm for Wreless Sensor Networks," Internatonal Journal of Future Generaton Communcaton and Networkng, vol.8, no. 1, pp. 109-116, 2015. [8] J. N. S.,and M. R. Mundada, "Localzaton technques for wreless sensor networks," Internatonal journal of engneerng and techncal research (IJETR), vol.3, no. 1, pp. 30-36, 2015. [9] Y. Shang, W. Ruml, Y. Zhang, and M. P. J. Fromherz, "Localzaton from mere connectvty," ACM symposum on moble Ad Hoc networkng and computng, pp. 201-212, 2003. [10] A. Pal, "Localzaton Algorthms n Wreless Sensor Networks: Current Approaches and Future Challenges," Network Protocols and Algorthms, vol. 2, no. 1, pp.45-73, 2010. [11] K. Langendoen, and N. Rejers, "Dstrbuted localzaton n wreless sensor networks: A quanttatve comparson," Computer networks, vol.43, no. 4, pp. 499-518, 2003. [12] P.D.Patl, and R.S.Patl, " Dstrbuted Localzaton n Wreless Ad-hoc Networks," IOSR Journal of Electroncs and Communcaton Engneerng (IOSR-JECE), Second Internatonal Conference on Emergng Trends n Engneerng,vol.2, pp. 34-38, 2013. [13] J. Borensten, and L. Feng, "Measurement and correcton of systematc odometry errors n moble robots," 186
IEEE Trans. Robot. Autom, vol. 12, no. 6, pp. 869 880, 1996. [14] P. Goel, S. I. Roumelots and G. S. Sukhatme, "Robot Localzaton Usng Relatve and Absolute Poston Estmates," Intellgent robots and systems, Internatonal conference, vol. 2, pp. 1134 1140, 1999. [15] J. Borensten, and L. Feng, "UMBmark A Method for Measurng, Comparng, and Correctng Dead-reckonng Errors n Moble Robots," The Unversty of Mchgan, Techncal Report UM-MEAM-94-22, December 1994. [16] W. Burgard, D. Fox, D. Henng and T. Schmdt, "Estmatng the Absolute Poston of a Moble Robot Usng Poston Probablty Grds," Proc. of the Fourteenth Natonal Conference on Artfcal Intellgence (AAAI-96). [17] B. Hofmann-Wellenhof, H. Lchtenegger, and J. Collns, " Global Postonng Systems: Theory and Practce," Sprnger, 5 edton, 2001. [18] C. ZHAO, Y. XU, and H. HUANG, " Sparse Localzaton wth a Moble Beacon Based on LU Decomposton n Wreless Sensor Networks," RADIOENGINEERING, vol. 24, no. 3, 2015. [19] P.BAHL, and V. N. PADMANABHAN, "RADAR: an n-buldng RF based user locaton and trackng system,". In Proceedngs of the 19thAnnual Jont Conference of IEEE Computer and Communcatons Socetes (INFOCOM), pp.775 784, 2000. [20] N.Bulusu, J.Hedemann, and D. Estrn. '' GPS-less low-cost outdoor localzaton for very small devces,'' IEEE Personal Communcatons, vol. 7,pp 28-34, 2000. [21] A. T. Rashd, M. Frasca, A. A. Al, A. Rzzo, and L. Fortuna, "Mult-robot localzaton and orentaton estmaton usng robotc cluster matchng algorthm," Robotcs and Autonomous Systems, vol. 63, pp. 108 121, 2015. 187