Coverage algorithms for visual sensor networks

Visual sensor networks (VSNs) are becoming increasingly popular in a number of application domains. A distinguishing characteristic of VSNs is to self-configure to minimize the need for operator control and to improve scalability. One of the areas of self-configuration is camera coverage control that is, how should cameras adjust their field-of-views to cover maximum targets? This is an NP-hard problem. We show that the existing heuristics have a number of weaknesses that influence both coverage and overhead. Therefore, we first propose a computationally efficient centralized heuristic that provides near-optimal coverage for small-scale networks. However, it requires significant communication and computation overhead, making it unsuitable for large-scale networks. Thus, we develop a distributed algorithm that outperforms the existing distributed algorithm with lower communication overhead, at the cost of coverage accuracy. We show that the proposed heuristics guarantee to cover at least half of the targets covered by the optimal solution. Finally, to gain benefits of both centralized and distributed algorithms, we propose a hierarchical algorithm where cameras are decomposed into neighborhoods that coordinate their coverage using an elected local coordinator. We observe that the hierarchical algorithm provides scalable near-optimal coverage with networking cost significantly less than that of centralized and distributed solutions.