On Target Coverage In Mobile Visual Sensor Networks

Recent advancements in manufacturing low-cost wireless battery operated cameras has made their application in Wireless Video Sensor Networks (WVSN) increasingly more feasible and affordable. The application of robotic sensing agents equipped with cameras in WVSNs, seems particularly promising in performing coverage tasks for ad hoc surveillance. Their application in this context can be specifically useful for surveying areas with little to no available or affordable infrastructure, or where quick deployment is necessary. In this paper, we address the target coverage problem for finding the minimum number of cameras, their placement, and orientation to cover arbitrarily located targets in an area of interest. We propose a computationally light-weight heuristic, where the number of used mobile cameras is close to those found by near-optimal algorithms. Specifically, we address this problem for non-uniform target distributions that naturally form clusters. Having light-weight heuristics will be particularly useful when the application is required to adapt to target mobility and/or is implemented in embedded systems. Our simulation study shows that when clusters are sufficiently separated, the required number of cameras found by our proposed method is very close to those acquired by the near-optimal algorithm, whereas the computational complexity of our algorithm is about ten times less. We also deploy our algorithm on a drone testbed using off-the-shelf components to verify its feasibility.