Sensor Selection for TOA-Based Multitarget Localization With Nonshared Sensors

In the multisensor localization problem, selecting sensors strategically to form an effective geometric configuration is crucial for enhancing both localization accuracy and system resource utilization. This study tackles the challenge of sensor selection for multitarget localization using time-of-arrival measurements. Distinguishing from previous works that focus on single-target scenarios with only shared sensors, multitarget scenarios with both nonshared and shared sensors within sensor networks are considered in this article. By adopting the A-optimal design criterion and introducing a Boolean matrix to indicate the sensor selection and allocation status, we formulate the sensor selection problem as an integer programming (IP) problem with nonconvex functional constraints imposed by nonshared sensors. Semidefinite relaxation technique is then employed to relax the rank-1 and integer constraints, transforming the IP problem into a semidefinite programming (SDP) problem. Further, a fast binary rounding algorithm is proposed to discretize the continuous solution of the SDP problem. Simulation results demonstrate that the proposed method significantly reduces the computational burden and provides superior performance compared to the exhaustive search method and other existing methods.