A Novel Map-Free Radio Indoor Localization Method Exploiting Multipath Spatial Consistency

Artikel i vetenskaplig tidskrift, 2026

Map-free indoor localization methods can achieve localization purpose without a prior environmental map, thus attracting interest in recent years. However, existing radio-based methods suffer from degraded localization accuracy in reflective non-line-of-sight (NLoS) environments. In this paper, we exploit multipath spatial consistency to develop a map-free radio indoor localization method. This method first constructs the map using spatial similarities among NLoS paths at known points, and then performs localization using the re-constructed map and the target NLoS components. Specifically, the proposed method consists of two stages: mapconstruction stage and target-localization stage. In the map-construction stage, first, given the angle-delay profile at known points (i.e., training points) within a region, the NLoS paths among all training points exhibiting spatial similarities are clustered; Then, the object associated with NLoS path cluster (i.e., shared object) is marked; Next, the non-convex problem of estimating locations of shared objects is formulated, and an iterative convex optimization (ICO)-based algorithm is proposed to obtain the suboptimal solution; Finally, the estimated results are used to construct the map (i.e., a map that includes shared object locations without specific dimensions). In the targetlocalization stage, given the re-constructed map and angle-delay profile at an unknown point (i.e., target point), the shared objects interacting with the target point are identified. The proposed ICO-based algorithm is then applied to solve the formulated non-convex problem of estimating the target location. The performance of our proposed method is validated through both radio propagation simulations and measurements, demonstrating localization performance in highly reflective environments.