Vehicle Localization Based on Hypothesis Test in NLOS Scenarios

Recently, vehicle-to-vehicle (V2V) based localization has attracted much attention due to its potential to achieve high accuracy. However, the error caused by non-line-of-sight (NLOS) propagation significantly affects the localization performance. In this paper, a novel method combining V2V communication and NLOS identification is proposed to improve accuracy of vehicle localization in NLOS scenarios. First, the algorithm identifies NLOS links with statistical methods. In this step, identification is conducted by decision theory or z-test, which depends on whether priori NLOS information is known or not. After that, NLOS links are discarded. Using the information including length of remaining links and GPS positions of surrounding vehicles, the estimated positions of target vehicles can be obtained by multilateration. It is shown by simulations that the proposed algorithm outperforms several classical methods in accuracy. Specifically, when priori NLOS information is available, 80% of vehicles have a localization error less than 5 m. For the case where NLOS information cannot be obtained, the algorithm still have good performance, and the ratio of vehicles having error within 5 m reaches 70%.