Surveillance with wireless sensor networks in obstruction: Breach paths as watershed contours
Introduction
Suppose that a rough terrain is to be monitored to detect unauthorized entries as shown in Fig. 1a. This task can be risky for humans. Instead of deploying a wired surveillance system, a wireless sensor network (WSN) is an easy alternative where the sensors can be dropped by an aircraft. As soon as the sensors configure themselves, they can start sensing the environment and communicate the target presence decisions to a sink.
Measures must be defined to analyze the sensing quality that signifies how well a surveillance wireless sensor network (SWSN) covers a region and senses the phenomena of interest such as an intrusion. The sensing quality depends on the characteristics of the sensor, target and environment, as well as the number and the deployment scheme of the sensors. The variety of sensor technologies makes the coverage analysis difficult because the underlying signal processing and detection procedures depend on the physics of the devices. If the target detection probability is well-defined, a quality of sensing measure can be established.
The sensing quality can be defined through how well the breach path is covered. Finding breach paths is referred to as the weakest breach path problem [1], or the best coverage problem [2]. Megerian et al. define the worst-and the best-case coverage of a WSN for homogeneous sensors [3]. The target wants to avoid the sensors. Thus, passing far from the sensor is good from the target’s viewpoint. The worst case coverage (maximal breach path) is defined with the quality of sensing measure as the closest distance to the sensors while the target crosses the field [4]. A similar problem derived from the worst case coverage definition is to find the farthest distance to the sensors when the target wants to stay as close as possible to the sensors. This problem is referred to as the best case coverage (maximal support path) problem. Such worst case measures can be employed in applications with high security requirements such as the surveillance of a mission-critical place. Clouqueur et al. consider the unauthorized traversal problem where the likelihood of detecting the target (referred as path exposure) is studied [5]. The field is modelled as a grid in [5] and the grid size is equal to the product of the target speed and the sensor sampling period. If the sampling periods of sensors are not synchronized, this approach weakens the accuracy. Moreover, signal characteristics and environmental conditions are not taken into account in [3], [5].
The main research question addressed in this paper is: How can the sensing quality be measured in the presence of obstacles in a SWSN? We propose a deployment (sensing) quality measure in terms of the detection probability of a target crossing the field on breach paths. This deployment quality measure (DQM) can be used to design the network to provide a required sensing quality and to answer typical questions such as how many sensors or how the sensors should be deployed. If the field of interest is known a priori, and deterministic deployment is possible, the designer may utilize the near optimal sensor placement algorithm proposed by Lin and Chiu [6] to provide a complete sensing coverage. For random deployment, it is concluded that the optimal sensor placement problem is NP-complete [6]. Younis and Akkaya present an extensive survey on the issues for node placement in WSNs in [7]. Therefore, it is practically important to have an accurate DQM because the main functionality of a SWSN is to secure a field-of-interest.
Section snippets
Breach paths
Traditionally, Voronoi decomposition is utilized to reveal breach paths based on exposure definitions [8], [3], [9], [10]. Voronoi decomposition of a discrete set of sensors distributed in the Euclidean space is the partition of the plane where each sensor is associated with a region. All of the points in the region of any sensor are closer to that sensor than any other. The borders between the regions, which are equidistant to multiple sensors, are the breach paths. Given a set S of discrete
Iso-sensing map definition
In this section, we define how the iso-sensing map is produced, and present how the watershed algorithm is used to determine the breach paths.
Watershed segmentation
In image processing, the gradient images which correspond to the topographic reliefs are often denoted with gray-scale pictures. The gray tones in the image depict the elevations in the region. Image segmentation is a process to discriminate the objects in an image from the background. A common approach is to find disjoint regions that are homogeneous with respect to some property. Watershed segmentation is a region-based approach and the idea behind this algorithm comes from nature. The
Deployment quality measure
The most secure path for a target follows the grid points that are the most distant from the sensors in the field. From the WSN’s point of view, this path is the weakest breach path. Thus, the maximum detection probability on the weakest breach path provides a measure to analyze the quality of deployment [1]. Watershed segmentation produces several contours in which the weakest breach path resides.
The watershed contours are the points in the iso-sensing map with the lowest detection
Simulation results and discussion
To analyze the watershed segmentation, we developed a simulator coded in C++ integrated with Matlab. In the simulations, a 300 × 60 m2 surveillance field is made up of 1 m × 1 m grids, and the boundary regions are 10 m wide. Sensors are deployed randomly with uniform distribution over the field. Thirteen obstacles are modeled as discs where the centers are also uniformly distributed and the radii are uniform random variables between 10 m and 20 m. The obstacles are assumed to disable sensing, as well as
Conclusion
In this paper, we propose the utilization of the watershed segmentation algorithm to find the possible breach paths in a surveillance field with obstacles. In order to apply the watershed segmentation, the iso-sensing map is defined and a recursive algorithm is designed to find the deployment quality measure defined as the maximum detection probability on the weakest breach path. The simulations indicate the impact of the false alarm rate and the sensor count on the deployment quality measure.
Acknowledgement
This work is supported by the State Planning Organization of Turkey under the Grant Number 03K120250, and by TUBITAK under the Grant Number 106E082. Parts of this work are presented in preliminary form in the IEEE International Conference on Communications, İstanbul, Turkey, June 2006.
Ertan Onur received the B.S. degree in Computer Engineering from Ege University, İzmir, Turkey in 1997, and the M.S. and Ph.D. degrees in Computer Engineering from Bogaziçi University, İstanbul, Turkey in 2001 and 2007, respectively. He is a BAL’93 graduate. After the B.S. degree, he worked for LMS Durability Technologies GmbH, Kaiserslautern, Germany and Global Bilgi, İstanbul, Turkey. During the Ph.D degree, he worked as a R&D project manager at Argela Technologies, İstanbul. Presently, he is
References (41)
- et al.
Strategies and techniques for node placement in wireless sensor networks: a survey
Ad Hoc Networks
(2008) - et al.
A delaunay triangulation based method for wireless sensor network deployment
Computer Communications
(2007) - et al.
Uncertainty-aware and coverage-oriented deployment for sensor networks
Journal of Parallel and Distributed Computing
(2004) - et al.
Energy-efficient coverage problems in wireless ad hoc sensor networks
Computer Communications
(2006) - et al.
Coverage and connectivity issues in wireless sensor networks: a survey
Pervasive and Mobile Computing
(2008) Topographic distance and watershed lines
Signal Processing
(1994)- et al.
How many sensors for an acceptable breach probability level?
Computer Communications
(2006) - et al.
Coverage in wireless ad hoc sensor networks
IEEE Transactions on Computers
(2003) - et al.
Worst and best-case coverage in sensor networks
IEEE Transactions on Mobile Computing
(2004) - et al.
Optimal coverage paths in ad hoc sensor networks
Sensor deployment strategy for detection of targets traversing a region
Mobile Networks and Applications
A near-optimal sensor placement algorithm to achieve complete coverage/discrimination in sensor networks
IEEE Communications Letters
Exposure in wireless sensor networks: theory and practical solutions
Wireless Networks
Redundancy and coverage detection in sensor networks
ACM Transactions on Sensor Networks
Voronoi diagrams – a survey of a fundamental geometric data structure
ACM Computing Surveys
Adaptive triangular deployment algorithm for unattended mobile sensor networks
IEEE Transactions on Computers
Watersheds in digital spaces: an efficient algorithm based on immersion simulations
IEEE Transactions on Pattern Analysis and Machine Intelligence
Cited by (0)
Ertan Onur received the B.S. degree in Computer Engineering from Ege University, İzmir, Turkey in 1997, and the M.S. and Ph.D. degrees in Computer Engineering from Bogaziçi University, İstanbul, Turkey in 2001 and 2007, respectively. He is a BAL’93 graduate. After the B.S. degree, he worked for LMS Durability Technologies GmbH, Kaiserslautern, Germany and Global Bilgi, İstanbul, Turkey. During the Ph.D degree, he worked as a R&D project manager at Argela Technologies, İstanbul. Presently, he is an assistant professor at Technical University of Delft, Netherlands. His research interests are in the area of telecommunications, personal networks, wireless and sensor networks. He is a member of IEEE.
Cem Ersoy received his B.S. and M.S. degrees in Electrical Engineering from Bogaziçi University, İstanbul, in 1984 and 1986, respectively. He worked as an R&D engineer in NETAS A.S. between 1984 and 1986. He received his Ph.D. in Electrical Engineering from Polytechnic University, Brooklyn, New York in 1992. Currently, he is a professor in the Computer Engineering Department of Bogˇaziçi University. His research interests include performance evaluation and topological design of communication networks, wireless networks and mobile applications. He is a senior member of IEEE.
Hakan Deliç received the B.S. degree (with honors) in Electrical and Electronics Engineering from Bogaziçi University, İstanbul, Turkey, in 1988, and the M.S. and the Ph.D. degrees in Electrical Engineering from the University of Virginia, Charlottesville, in 1990 and 1992, respectively. He was a Research Associate with the University of Virginia Health Sciences Center from 1992 to 1994. In September 1994, he joined the University of Louisiana at Lafayette, where he was on the Faculty of the Department of Electrical and Computer Engineering until February 1996. He was a Visiting Associate Professor in the Department of Electrical and Computer Engineering, University of Minnesota, Minneapolis, during the 2001–2002 academic year. He is currently a professor of Electrical and Electronics Engineering at Bogˇaziçi University. His research interests lie in the areas of communications and signal processing, with current focus on ultra-wideband communications, iterative decoding, robust systems, and sensor networks. He frequently serves as a consultant to the telecommunications industry. He is a senior member of IEEE.
Lale Akarun received the B.S. and M.S. degrees in Electrical Engineering from Bogaziçi University, İstanbul, in 1984 and 1986, respectively. She obtained her Ph.D. from Polytechnic University, New York in 1992. Since 1993, she has been working as a faculty member at Bogˇaziçi University. She became a professor of Computer Engineering in 2001. Her research areas are face recognition, modeling and animation and human activity and gesture analysis. She has worked on the organization committees of IEEE NSIP99, EUSIPCO 2005, and eNTER-FACE2007. She is a senior member of the IEEE.