An Efficient Sensor Selection Algorithm for TDOA Localization with Estimated Source Position

This paper focuses on improving the sensor selection performance in the time difference of arrival (TDOA)-based localization scenario with the presence of source estimation error. In existing schemes, a coarse source position is first estimated and regarded as the actual counterpart for selecting optimal sensors. However, if the estimated position deviates from the actual position, the localization accuracy determined by the selected sensor subset will severely degrade. To solve the issue, we devise a distance-related weighted average Cramer Rao lower bound (WA-CRLB) to include the spatial information of the actual position by scattering sampling points around the estimated source position according to its distribution. Then, we formulate a Boolean vector-based sensor selection optimization problem to minimize WA-CRLB and propose a modified iterative swapping greedy (MISG) algorithm. Simulation results show that the proposed MISG algorithm achieves higher localization robustness with the increase of TDOA measurement error strength, and has lower computational complexity compared with the previous semi-definite relaxation (SDR)-based algorithms.