Optimization framework for battery-powered sensor selection for sound source detection

This study aims to address the multi-objective problem of selecting the optimal sensors to activate from a spatially distributed sensor array for sound source localization. We aim to achieve this with minimal sensor deployment, thereby maximizing the battery life of a device that powers both the sensors and an onboard processing system. The optimization problem has two conflicting objectives: (a) minimize the sensor deployment, which directly translates to lower power consumption and (b) maximize source localization confidence. We present a novel deterministic greedy algorithm for solving the multi-objective optimization problem. A dictionary of optimized sensor locations is first pre-computed and stored for a variety of scenarios for single source locations case. For test cases, we use the dictionary to generate an initial guess for the actual greedy optimization algorithm. We demonstrate the efficiency of the algorithm through various simulated test cases.