The non-line of sight problem in mobile location estimation

Abstract

We consider the problem of tracking mobile stations using the ranging measurements from multiple base stations (BSs), without a priori knowledge of which BSs (if any) have a direct line of sight (LOS) range measurement. The two key contributions of this paper are to show that (1) it is possible to discriminate between LOS versus non-LOS (NLOS) measurements at each BS by using the time history of its range measurements in a simple hypothesis testing problem and (2) it is possible to correct the NLOS ranging error by exploiting a priori knowledge of the statistical characteristics of the system's standard measurement noise. Simulation examples are presented and the RMS error is compared to the Cramer Rao lower bound on location estimation.