Computationally Efficient Target-Node Geometry Selection For Target Tracking In Uwsns

Since underwater nodes provide measurements for target tracking based on underwater wireless sensor networks (UWSNs), target-node geometry (T-NG) may affect the performance of target tracking. This paper studies T-NG effect on target tracking in UWSNs using quantized measurements. In order to evaluate the arbitrary T-NG, the relation between posterior Cramer-Rao lower bound (PCRLB) and node's position is derived. In general, an exhaustive search is required to find the optimal T-NG by minimizing PCRLB, which is computationally prohibitive and not realistic for real-time online implementation. To decrease computational complexity while keeping proper tracking accuracy, this paper utilizes the generalized Breiman, Friedman, OIshen, and Stone (GBFOS) algorithm, the greedy search, and the random scheme to select the best T-NG. Their performance is compared with the exhaustive search. Although the random scheme is the fastest, its tracking error is too large. The greedy search and GBFOS can decrease computational load greatly while keeping almost the same tracking accuracy as that of the exhaustive search. Simulations are carried out in terms of the dense network and the sparse network, and the results can demonstrate the effectiveness of the greedy search and GBFOS, especially when the network is relatively dense.