| Summary: | Barrier coverage is one of the most important issues in intruder detection applications. Visual sensors can give more accurate information of intruder but also raise the problem of large data volume for information exchange and processing. Constructing a disjoint <italic>full-view</italic> barrier using minimal number of visual sensors has been the main goal for handling the barrier coverage problem. This paper presents a decentralized FBCA mechanism which consists of <italic>Region Partitioning Phase</italic>, <italic>Grid Excluding Phase</italic>, <italic>Grid Verification Phase</italic> and <italic>Full-View Barrier Construction Phase</italic>. In <italic>Region Partitioning Phase</italic>, the given <inline-formula> <tex-math notation="LaTeX">$R $ </tex-math></inline-formula> is partitioned into a set of equal-sized grids, aiming to simplify the construction problem of the full-view barrier. In <italic>Grid Excluding Phase</italic>, a certain amount of grids is removed, aiming to reduce the computational complexity. In <italic>Grid Verification Phase</italic>, each visual sensor aims to check if its neighboring grids satisfy the full-view coverage criteria. Finally, the <italic>Full-View Barrier Construction Phase</italic> aims to construct as many as possible full-view barriers. Experimental study shows that the proposed FBCA outperforms existing work SP, and likely approaches to the optimal solution (MCSPS) in terms of the average success ratio for constructing a <italic>full-view</italic> barrier.
|
|---|