Multi-sensor system deployment planning method for underwater surveillance based on formation characteristics

Abstract

The deployment planning issue for a multi-sensor system comprising a limited number of sensors designed to detect underwater intrusion targets is defined as a multi-objective NP-hard problem. This problem is constituted by two competing and incommensurable optimization objectives: "larger sensor coverage" and "higher probability of detecting intrusion targets". The map of the mission area is transformed into a topological map through the application of polygon fitting and segmentation based on Delaunay triangulation. This study employs a characteristics-based non-dominated sorting genetic algorithm (CBNSGA) to address the deployment planning issue of the multi-sensor system. In this algorithm, Mean-Shift clustering is employed to yield characteristics information through the clustering of the multi-sensor system formation. Subsequently, this information is employed to enhance the crossover, mutation, and selection strategies. Adaptive parameters are designed to accelerate convergence and avoid local optima. Additionally, the Cauchy inverse cumulative distribution operator is employed to enhance the mutation step. The feasibility and effectiveness of the CBNSGA in multi-sensor system deployment planning are demonstrated through simulation and comparison with other algorithms.