Discrete position processing techniques for indoor NLOS localization

In this paper we investigate localization in indoor scenarios which are characterized by non-line-of-sight multipath propagation and rich scattering. In such environments, classical position and trajectory estimation methods yield rather unreliable position estimates. We present a technique to estimate the most likely position or the most likely trajectory of a mobile terminal given a sequence of observations. Discrete positions are considered for computational feasibility of the processing techniques. The most likely position of the mobile terminal is estimated using the forward-backward algorithm whereas the most likely trajectory of the mobile terminal is estimated using the Viterbi algorithm. The observations are the range, the angle-of-arrival and the angle-of-departure. The mobility model of the mobile terminal is defined by a Markov model. It is shown, through simulations, that the proposed processing methods yield an increase in performance compared with classical least squares and maximum likelihood position estimation methods. In addition, the performance versus complexity trade-off is also discussed.