Poster Abstract: A Distributed Algorithm to Compute Spatial Skyline in Wireless Sensor Networks * SunHee Yoon and Cyrus Shahabi Department of Computer Science, University of Southern California sunheeyo@usc.edu, shahabi@usc.edu ABSTRACT Spatial skyline queries can be used in wireless sensor networks for collaborative positioning of multiple ob- jects. However, designing a distributed spatial skyline algorithm in resource constrained wireless environments introduces several research challenges: how to combine multi-dimensional data, (e.g. distances to multiple events) to compute the skylines efficiently, accurately, quickly, progressively, and concurrently while dealing with the network and event dynamics. We address this challenge by designing Distributed Spatial Skyline (DSS) algo- rithm. DSS is the first distributed algorithm to com- pute spatial skylines. In a network of 554 nodes, DSS reduces the communication overhead by up to 91% over a centralized algorithm with 100% accuracy. 1. INTRODUCTION Given a set of data points P and a set of query points Q, spatial skylines are those members of P that are not spatially dominated by any other point in P with re- spect to Q [2]. A point p1 spatially dominates a point p2 with respect to Q, if and only if p1 is closer to at least one query point as compared to p2 and has the same distance as p2 to the rest of the query points. The spatial skyline query can be utilized to collaboratively position multiple targets in wireless networks. For ex- ample, suppose from a set of police cars (data points P ) we want to identify a candidate subset to be dispatched to multiple incident points (query points Q). This can- didate subset includes those cars that are not dominated by any other police car with respect to all the incidents, * This research has been funded in part by NSF grants IIS-0238560 (PECASE) and CNS-0831505 (CyberTrust), the NSF Center for Embedded Networked Sensing (CCR- 0120778) and in part from the METRANS Transportation Center, under grants from USDOT and Caltrans. Any opinions, findings, and conclusions or recommendations ex- pressed in this material are those of the author(s) and do not necessarily reflect the views of the National Science Founda- tion. Copyright 2009 ACM 978-1-60558-371-6/09/04 ...$5.00. GHT-based routing  10 4 16 B A C 1 15 11 12 13 14 9 0 3 6 5 2 8 E 11 1 0 2 3 5 6 7 7 8 9 16 12 13 14 15 A B C D E 10 R (a) Phase 1: Computing the query convex hull and R. C D 6 3 0 1 9 2 11 13 12 8 A 5 14 B A C 10 4 16 7 5 15 11 1 4 7 8 10 12 13 16 15 D B 14 3 9 0 6 2 A 5 C 16 8  7 7 R (b) Phase 2 and 3: Search internal, external skylines. Figure 1: DSS algorithm. {A,B,C,D,E} are query points and quadrilateral ABDC is the query convex hull. Star is a synchronization point, R is a rendezvous point, and sunflowers are spatial skylines {0,8,14,2,3,9,6,5,16}. and hence they are the spatial skylines. Spatial sky- line query can be performed using either centralized or distributed approaches. The centralized approach assumes a complete set of data in space and time; if we use TAG-like data collec- tion tree [1], it requires high transmission overhead for resource constrained wireless networks. For the time- critical applications, centralized approach cannot meet the promptness requirement because it does not provide any partial results early. We address these challenges by proposing a efficient Distributed Spatial Skyline (DSS) algorithm to collaboratively find the spatial skylines in WSN. DSS parallelizes the search for skylines by parti- tioning the search space recursively based on the geo- metric properties of the nodes and the topology. This approach enables an efficient, scalable, progressive, fast, and concurrent search while providing correct results. Our DSS algorithm is the first distributed algorithm to compute spatial skylines. 2. DSS ALGORITHM The DSS algorithm partitions the network into two sets of nodes: nodes inside or outside of query convex hull CH(Q) which is the convex hull of query points Q.