Geometric Algorithms for Coverage in Wireless Sensor Networks Dr. Dinesh Dash Asst. Prof.NIT Patna Outline of Talk Introduction Application of geometric algorithms in Sensor Network Geometric Algorithm for Coverage problem in sensor network Area Coverage Barrier Coverage Line Coverage Introduction Sensor networks composed of a large number of sensor nodes, which are deployed to sense environmental parameters and send it to a sink. Example of Sensor Network Sensing devices sense environment, do some local processing and send the sensed data to a base station directly or indirectly. collator collator Base station collator Some Applications Sensor networks have been used for habitat monitoring [mainwaring:2002], agriculture monitoring [bilsa:2009], structure monitoring, forest fire detection, object tracking [tsaia:2007], military application, etc. Applications of Geometric Algorithms in Sensor Network Routing Localization Coverage Location Based Routing Assumption source Every node knows its location and its neighbors’ locations The source knows the location of destination Greedy Forwarding A node always forwards the message to a neighbor whose Euclidean distance to the destination is smaller destination Greedy forwarding may fail The message reaches node x, no next hop can be selected for Greedy Forwarding, because both w and y are further away from D than x is. Planer Graph and Face Routing helps Localized ways to planerize a unit disk graph Gabriel Graph Relative Neighborhood Graph Face Routing face face face face face source destination Localization Determines physical / relative positions of sensor nodes in the network based on known information Essential for: The development of low-cost sensor networks for use in location-aware applications Geographic routing Beacon Based Algorithm Coverage Coverage problem measures how well a set of deployed sensors cover(sense) an area or a set of objects Different measures of coverage is possible depending on its applications Area coverage, target coverage, barrier coverage, trap coverage, perimeter coverage, line coverage, breach and support etc. Coverage Algorithms Measures of Coverage: Examples Area coverage [thai:2008, bai:2005, huang:2005] Measures coverage of an area/region Can be full coverage (every point of the area is sensed by at least one sensor) or partial (some fraction of the area is covered) Target coverage [cardei :2005] Given a set of target points, each point in the set is sensed by at least one (or k) sensor Area Coverage Is the entire area covered? 1 8 2 3 4 6 5 Is the entire region k-covered? An area A is k-covered if all intersection points among the sensing circles and area boundaries are k-covered There are at most O(n2) intersection points Each intersection point can be verify in O(n) An area A is k-covered iff each sensor in A is k perimeter covered 1 Is the perimeter k-covered? 5 7 6 8 4 Perimeter coverage is modeled by set of intervals and can be verified in O(n log n ) 2 3 9 4 0 5 3 1 2 6 5 4 2П 10 Barrier coverage A rectangular belt region is said to be 1-barrier covered by the deployed sensors if all the crossing paths must intersect at least one sensor’s sensing region [kumar:2005] forts were surrounded by deep trenches Barrier formed by sensors L R Open barrier Closed barrier Graph view of barriers L R One vertex for each sensor, two dummy vertices L and R for left and right boundaries in open barrier Edge between two vertices if the sensing regions of the two sensors intersect For open barrier a path from L to R in the graph ensures barrier coverage For closed barrier a non-contractible cycle ensures barrier coverage Is a boundary k-barrier covered? Construct a graph G(V, E) V: sensor nodes, plus two dummy nodes L, R E: edge (u,v) if their sensing disks overlap Region is k-barrier covered iff there are k-disjoint paths between L and R L R Line/path coverage Measures the degree of coverage of lines/paths Some variants Every point of a line segment is under the sensing range of at least one sensor [harada:2009] fraction of coverage [harada:2009] : the fraction of the whole path that is within the sensing range of some sensors Path Coverage Maximal breach and support path [megerian:2005 ] Breach path A path that want to maintain distance from the sensors Support path A path that want to stay close to the sensors Breach and Support i Breach value of a path is the minimum distance of any point on the path from the closest sensor Support value of a path is the maximum distance of any point on the path to its closest sensor Breach value 18 5 Pi Pf support value f a) The maximal breach path -> Voronoi diagram b) The maximal support path -> Delaunay triangulation How to find maximal breach path? Theorem : At least one Maximal Breach Path must lie on the line segments of the bounded Voronoi diagram formed by the locations of the sensors. Apply binary search and breadth first search on the weighted edge of the Voronoi diagram to find a maximal breach path How to find maximal support path? Theorem : At least one maximal Support Path must lie on the edges of the Delaunay triangulation. Apply binary search and breadth first search on the weighted edge of Delaunay triangulation to find a maximal support path Find Euclidean minimum spanning tree of the set of sensors including initial and final points. Line Coverage Problem A line segment is said to be k-covered if it is sensed by k sensors 3-covered line segment 3-uncovered line segment Definition Smallest k-covered line segment Given a sensor deployment, the minimum length line segment that intersects at least k sensors’ sensing regions Longest k-uncovered line segment Given a sensor deployment, the maximum length line segment that intersects at most k-1 sensors’ sensing regions Maximal 3-uncovered line segment Minimal 3-covered line segment Problems Addressed Designed algorithms for finding smallest k-covered and longest kuncovered line segment given a sensor deployment in a bounded rectangular region R for Axis-parallel line segments in O((n+ χ) log n) time and linear space, where n is the number of sensors and χ is the number of intersections between the circles corresponding to the sensor’s sensing regions Arbitrary line segment starting from a given point in O((n+χ) log n) time and linear space Arbitrary line segments Smallest k-covered segment in O(χ2logn+n4) time and linear space Longest k-uncovered segment in O((χ2+n3)logn) time and linear space JPDC 2014 Algorithm for Smallest k-Covered Axis-Parallel Line Segment Overall approach Find the smallest k-covered horizontal line segment Find the smallest k-covered vertical line segment Choose the one with the minimum length among the two Only the algorithm for horizontal line segments discussed, the algorithm for vertical line segments is the same Some Definitions Top end point Left-half circle Right-half circle Bottom-end point Left-left intersection Left-right intersection Right-right intersection Note: Two equal radius circles can make only one left-left intersection and one right-right intersection Minimal length k-covered horizontal segment: A k-covered horizontal line segment/interval such that no subinterval of it is k-covered. Minimal length 3-covered Horizontal segment Non Minimal length 3-covered Horizontal segment Midpoint between Two Circles Midpoint between two circles Cp and Cq is the ycoordinate where the horizontal distance between the two circles is minimum. For unit circle, it is at (yp+yq)/2 where yp and yq are the y–coordinates of the centers of Cp and Cq respectively (xq,yq) (xp,yp) ymid = (yp+yq) /2 Approach Use plane sweep paradigm Horizontal sweep line moves from top to bottom of region Proved that minimal length k-covered horizontal line segments are only created or deleted when the sweep line touches a top-endpoint, a bottomendpoint, or a left-left or right-right intersection Keep track of all such minimal length k-covered horizontal segments created Determine the minimum length attained by each segment (using the mid-point events) Update the global minimum as necessary Sweep-Line Touches the TopEndpoint of a Circle Sweep line c0 r l [c0r,c2l] [c1 ,c4 ] c1 [c1r,c3l] c2 c4 [c2r,c5l] [c2r,c4l] c3 [c3r,c5l] [c4r,c6l] c6 c5 Before touching top-endpoint of c3 , the set of minimal length 3-covered segments, are [c0r,c2l], [c1r,c4l], [c2r,c5l], [c4r,c6l] As it touches top-endpoint of c3 Existing segments [c1r,c4l], [c2r,c5l] are deleted New segments [c1r,c3l], [c2r,c4l], [c3r,c5l] are created Lemma : When the sweep-line touches the top-endpoint of a circle, at most k new minimal length k-covered segments can be created on the sweep line. Moreover, if x (x>0) k-covered segments are created, then exactly x−1 existing k-covered segments will be deleted. Sweep-Line Touches the BottomEndpoint of a Circle Sweep line c0 c1 [c3r,c5l] [c2r,c4l] [c1r,c3l] c2 [c2r,c5l] c4 [c4r,c6l] c6 c5 c3 [c3r,c6l] Before touching bottom-endpoint of c4 , the set of minimal length 3-covered segments, are [c1r,c3l] , [c2r,c4l], [c3r,c5l] , [c4r,c6l] As it touches the bottom-endpoint of c4 Existing segments [c2r,c4l], [c3r,c5l] , [c4r,c6l] are deleted New segments [c2r,c5l], [c3r,c6l] are created Lemma : When the sweep-line touches the bottom-endpoint of a circle, at most k existing k-covered segments are deleted from the sweep line. Moreover, if x (x>0) existing kcovered segments are deleted, then exactly x−1 new k-covered segments will be created. Sweep-Line Crosses Right-Right Intersection of Two Circles Sweep line c0 [c0r,c2l] c1 [c2r,c3l] [c1r,c4l] [c1r,c3l] c2 [c3r,c5l] c3 c4 c5 [c2r,c4l] Right-right intersection Before touching right-right intersection between c1 & c2 , the set of minimal length 3-covered segments are [c0r,c2l] , [c2r,c3l], [c1r,c4l] , [c3r,c5l] As it crosses the right-right intersection between c1 & c2 Existing segments [c2r,c3l] & [c1r,c4l] are deleted New segments [c1r,c3l] & [c2r,c4l] are created Sweep-Line Touches Left-Left or Right-Right Intersection Lemma : When the sweep-line touches the leftleft or right-right intersection point between two circles then it creates at most two new k-covered segments. Moreover, if x (0 ≤ x ≤ 2) new kcovered segments are created, then exactly x existing k-covered segments are deleted. Sweep-line touches left-right intersection of two circles Sweep line c0 [c0r,c2l] [c0r,c2l] c2 c1 [c1r,c3l] [c1r,c3l] c3 left-right intersection Before touching left-right intersection between c1 & c2 , the set of minimal length 3-covered segments are [c0r,c2l] , [c1r,c3l] As it touches the intersection no existing segments are deleted as well as no new segments are created When sweep-line crosses any left-right intersection point no new kcovered intervals are created or deleted Algorithm Outline Move the sweep line from top to bottom Identify when it touches a top-endpoint, bottomendpoint, left-left intersection, right-right intersection or the mid-point where the length of a k-covered line segment is minimized (events) Keep track of the minimal length k-covered segments formed, deleted, and their minimum lengths on each such event References [thai:2008] M. T. Thai, F. Wang, and D. Du. “Coverage problems in wireless sensor networks: Designs and analysis”, ACM Journal on Sensor Network, 3(3):191-200, 2008. [megerian:2005] S. Megerian, F. Koushanfar, M. Potkonjak, and M. B. Srivastava. “Worst and best-case coverage in sensor networks”, IEEE Trans. on Mobile Computing, 4(1):753-763, 2005 [kumar:2005] S. Kumar, T. H. Lai, and A. Arora. “Barrier coverage with wireless sensors”, In ACM MOBICOM, pages 284-298, Aug. 2005 [harada:2009] J. Harada, S. Shioda, and H. Saito. “Path coverage properties of randomly deployed sensors with finite data-transmission ranges”, Computer Networks, 53(7):1014-1026, 2009. [cardei:2005] M. Cardei, M.T. Thai, Y. Li, and W. Wu. “Energy efficient target coverage in wireless sensor networks”, In IEEE INFOCOM, page 1976-1984, March 2005. [saipulla:2010] A. Saipulla, B. Liu, G. Xing, X. Fu, and J. Wang. “Barrier coverage with sensors of limited mobility”, In Intl. Symp. on Mobile Ad Hoc Networking and Computing, pages 201-210, Sept. 2010. [wu:2007] C. H. Wu, K. C. Lee, and Y. C. Chung. “A delaunay triangulation based method for wireless sensor network deployment”, ACM Computer Communications, 30(14-15), 2007. [chakrabarty:2002] K. Chakrabarty, S. S. Iyengar, H. Qi, and E. Cho. “Grid coverage for surveillance and target location in distributed sensor networks”, IEEE Trans. on Computers, 51(12), 2002. [agnetis:2009] A. Agnetis, E. Grande, P. B. Mirchandani, and A. Pacifici. “Covering a line segment with variable radius discs”, Computers and Operations Research, 36(5), 2009. [sekhar:2005] A. Sekhar, B. S. Manoj, and C. S. R. Murthy. “Dynamic coverage maintenance algorithms for sensor networks with limited mobility”, In Intl. Conf. on Pervasive Computing and Communications, pages 51–60, March 2005 [wang:2005] G. Wang, G. Cao, T. L. Porta, and W. Zhang. Sensor relocation in mobile sensor networks. In IEEE INFOCOM, pages 2302–2312, March 2005. [huang:2005] C. Huang and Y. Tseng. “The coverage problem in wireless sensor network”, Mobile Networks and Applications, 10(4), 2005. [tsaia:2007] H.W Tsaia, C.P. Chua, T.S. Chenb; “Mobile object tracking in wireless sensor networks”, Computer Communications Vol. 30, No. 8, 2007 [zhao:2009] M.C. Zhao, J. Lei, M.Y. Wu, Y. Liu, and W. Shu. “Surface coverage in wireless sensor networks”, In IEEE INFOCOM, pages 109–117, April 2009. [ram:2007] S. S. Ram, D. Manjunath, S. K. Iyer, and D. Yogeshwaran. On the path coverage properties of random sensor networks. IEEE Trans. on Mobile Computing, 6(5), 2007. [polastre:2004] J. Polastre, R. Szewczyk, C. Sharp, and D. Culler. The mote revolution: Low power wireless sensor network devices. In Hot Chips, 2004. [bai:2005] H. Bai, X. Chen, Y.-C. Ho, and X. Guan. Percentage coverage configuration in wireless sensor networks. In Intl. Conf. on Parallel and Distributed Processing and Applications, pages 780–791, Nov. 2005. [bilas2009] Vedran Bilas, Faculty of Electrical Engineering andComputing Wireless Sensor Networks in Agriculture – Olive Growing; University of Zagreb,2009 [balister:2009] P. Balister, Z. Zheng, S. Kumar, and P. Sinha. Trap coverage: Allowing coverage holes of bounded diameter in wireless sensor network. In IEEE INFOCOM, pages 136–144, April 2009 [bhattacharya:2009] B. Bhattacharya, M. Burmester, Y. Hua, E. Kranakis, Q. Shi, and A.Wiesed. “Optimal movement of mobile sensors for barrier coverage of a planar region”, Theoretical Computer Science, 410(52), 2009 Thanks!