Optimization of Source and Receiver Placement in Multistatic Sonar Environments.

摘要

The usage of non-collocated sources and receivers in multistatic sonar networks (MSNs) has emerged as a promising area of opportunity in sonar systems. This thesis addresses point coverage sensing problems in MSNs, where a number of points of interest (targets) must be monitored. For detection we assume cookie-cutter sensors and a two-dimensional homogenous environment. Based on current algorithms and the theory of detection discs, we develop the new preprocessing algorithm LOC-GEN-II to determine possible locations for sources given a set of targets and receivers. The high efficiency of this algorithm is based on the greedy-like approach it is built upon and allows a significant reduction of computing time compared to a recent algorithm from the literature. We also address the problem of optimally placing multiple sources and receivers for a given set of targets. Up to now, this problem was solved with the simplification of setting receivers randomly and placing only sources optimally. We develop LOC-GEN-II further into a two-step process of determining near optimal positions for receivers and sources successively. The procedure is implemented as a faster one-step solution and a slower iterative approach, which leads to better detection results. With this approach we show that the newly developed algorithms allow solution of multiple sensor placement problems in an acceptable time with significantly better detection results compared to the benchmark.