Environment Topology Oriented Fingerprint Localization in Indoor NLoS Scenarios

Previous studies in fingerprint localization have primarily used feature comparison for user equipments’ (UEs’) position estimation. The estimation accuracies heavily rely on the constructed feature and its sampled resolution. This work presents an environment-oriented fingerprint localization scheme in an indoor non-line-of-sight (NLoS) scenario, which first finds the locations of scatters that reflect signals from the base station (BS) to UEs. Given the locations of BS and scatters, we match each scatter with the estimated time-of-arrivals (ToAs) of the BS-scatter-UE links. This allows us to compute the ToAs from the scatters to UEs and convert the NLoS localization problem into a line-of-sight (LoS) one. Simulation results demonstrate that we can estimate the scatters’ positions with centimeter-level accuracy and about 90% of 1000 uniformly distributed UEs have localization errors less than 0.8m.