A Bayesian Probabilistic Approach to Hybrid Localization with GNSS and LTE-OTDOA in Multipath Channels

For time-of-arrival (TOA) localization, the channel bias introduced by unresolvable multipath and non-line-of-sight (NLOS) reflections severely degrade the performance. To address this impairment, Perez-Cruz et al. proposed a Bayesian probabilistic approach, which characterizes the channel bias with a probability distribution such that it can be robustly compensated, and illustrated its effectiveness for Long Term Evolution (LTE) - observed time difference of arrival (OTDOA) positioning. In this work, we generalize this Bayesian probabilisitic approach to hybrid positioning with both global navigation satellite system (GNSS) and LTE-OTDOA. Based on actual over-the-air measurement data, we demonstrate that the proposed probabilisitic hybrid localization algorithm achieves better positioning accuracy compared with the original one using GNSS or LTE-OTDOA observations only. It also significantly outperforms a baseline nonlinear hybrid multilateration algorithm, which does not take into account the channel bias error.