Pedestrian tracking algorithm in NLOS environments

Abstract

This paper presents a cellular network based positioning algorithm in an urban environment characterized by multipath and severe non line of sight (NLOS) errors. The proposed algorithm mitigates localization error up to 75% as shown by the simulation results. The algorithm involves an initial averaging step followed by a prediction step for optimization, confining the estimated location close to the actual location. The proposed algorithm doesn't require additional hardware like sensors, accelerometers, gyroscopes etc. for localization as used in traditional cellular network based positioning methods. This approach can also be utilized in indoor positioning system (IPS) and global positioning systems (GPS) when at most three satellites are available. Low computational complexity of the algorithm is an added advantage. Utilization of orthogonal sources of information for improving accuracy is also explored.