Heuristic algorithm for finding boundary cycles in location-free low density wireless sensor networks Lanny Sitanayah * , Amitava Datta, Rachel Cardell-Oliver School of Computer Science and Software Engineering, The University of Western Australia, Perth WA 6009, Australia article info Article history: Received 13 November 2008 Received in revised form 21 October 2009 Accepted 17 January 2010 Available online 25 January 2010 Responsible Editor: X.S. Shen Keywords: Wireless sensor networks Boundary cycles Location-free Distributed algorithm abstract Wireless sensor networks (WSNs) comprise a large number of sensor nodes, which are spread out within a region to be monitored and communicate using wireless links. In some WSN applications, recognizing boundary nodes is important for topology discovery, geo- graphic routing, tracking and guiding. In this paper, we study the problem of identifying the boundary nodes of a WSN. In a WSN, close-by nodes can establish direct communica- tions with their neighbors and have the ability to estimate distances to nearby nodes, but not necessarily the true distances. Our objective is to find the boundary nodes by using only the connectivity relation and neighbor distance information without any other knowl- edge of node locations. Moreover, our main aim is to design a distributed algorithm that works even when the average degree is low. We propose a heuristic algorithm to find the boundary nodes which are connected in a boundary cycle of a location-free, low den- sity (average degree 5–6), randomly deployed WSN. We develop the key ideas of our boundary detection algorithm in the centralized scenario and extend these ideas to the dis- tributed scenario. The distributed implementation is more realistic for real WSNs, espe- cially for sparse networks when all local information cannot be collected very well due to sparse connectivity. In addition, the distributed implementation can tolerate faults by recomputing the boundary locally when a boundary node is faulty. Simulations in ns-2 show that the distributed implementation outperforms the centralized one with higher quality of boundaries. Ó 2010 Elsevier B.V. All rights reserved. 1. Introduction Rapid improvement in wireless communication and electronics technologies have enabled the development of small, low-cost, low-power, multifunctional devices, sen- sor nodes. A sensor node (or mote) is a battery-powered device with integrated sensing, processing and communi- cation capabilities. It can detect and monitor changes in a variety of physical conditions, such as temperature, humidity, light, sound, chemicals, or the presence of cer- tain objects [17]. Nodes can perform simple computations and communicate with each other over short distances using radio. A wireless sensor network (WSN) is composed of hundreds to thousands of unattended sensor nodes and one or more base stations. The sensor nodes are deployed either densely or sparsely, manually or randomly, in a re- gion to be monitored, for example, natural environments, battlefields, hospitals, houses and industries. The arrange- ment and management of a WSN depends on the applica- tion for which it is used, such as military, environmental, health, home and some commercial applications [2]. In some WSN applications, recognizing boundary nodes is necessary for topology discovery [11–13,18], geographic routing [6,13], tracking and guiding [13]. In the applica- tions that involve tracking moving objects, such as in mil- itary applications for tracking enemy vehicles and detecting illegal border crossings, and in environmental applications for habitat monitoring [14], boundary 1389-1286/$ - see front matter Ó 2010 Elsevier B.V. All rights reserved. doi:10.1016/j.comnet.2010.01.009 * Corresponding author. Tel.: +61 411 782 558. E-mail addresses: lanny@csse.uwa.edu.au (L. Sitanayah), datta@csse. uwa.edu.au (A. Datta), rachel@csse.uwa.edu.au (R. Cardell-Oliver). Computer Networks 54 (2010) 1630–1645 Contents lists available at ScienceDirect Computer Networks journal homepage: www.elsevier.com/locate/comnet