A Min-Max Approximation-Based SDP Approach to Robust Vehicle Location Estimation in VANETS

In this paper, we consider robust vehicle localization in a vehicular ad-hoc network (VANET) under changing vehicle velocities and probabilistically mixed line-of-sight (LOS)/non-LOS (NLOS) conditions, where NLOS error statistics and occurrence probabilities are unknown. To tackle the problem, we develop two localization algorithms based on a proposed min-max approximation (MMA)-based semidefinite programming (SDP) method and a proposed advanced measurement preprocessing (AMP) scheme. The AMP method is used for reducing the NLOS influence on raw range measurements received by multiple base stations. Accordingly, we have AMP-MMA and AMP-Joint MMA algorithms. The latter can perform better than the former because it additionally incorporates a neighbor-assisting localization scheme. Simulations show that the two proposed location estimation algorithms perform satisfactorily and outperform the MMA and some tracking algorithms in the literature.