Adaptive Multipath-Based SLAM for Distributed MIMO Systems
- Published
- Tue, Jul 01, 2025
- Tags
- rotm
- Contact
Localizing users and mapping the environment using radio signals is a key task in emerging applications such as reliable, low-latency communications, location-aware security, and safety-critical navigation. Recently introduced multipath-based simultaneous localization and mapping (MP-SLAM) can jointly localize a mobile agent and the reflective surfaces in radio frequency (RF) environments. Most existing MP-SLAM methods assume that map features and their corresponding RF propagation paths are statistically independent. These existing methods neglect inherent dependencies that arise when a single reflective surface contributes to different propagation paths or when an agent communicates with more than one base station (BS). Previous approaches that aim to fuse information across propagation paths are limited by their inability to perform ray tracing in RF environments with nonconvex geometries.
In this paper, we propose a Bayesian MP-SLAM method for distributed MIMO systems that addresses this limitation. In particular, we make use of amplitude statistics to establish adaptive time-varying detection probabilities. Based on the resulting “soft” ray-tracing strategy, our method can fuse information across propagation paths in RF environments with nonconvex geometries. A Bayesian estimation method for the joint estimation of map features and agent position is established by applying the message passing rules of the sum-product algorithm (SPA) to the factor graph (FG) that represents the proposed statistical model. We also introduce an improved proposal PDF for particle-based computation of SPA messages. This proposal PDF enables the early detection of new surfaces that are solely supported by double-bounce paths. Our method is validated using synthetic RF measurements in a challenging scenario with nonconvex geometries. The presented results demonstrate that it can provide accurate localization and mapping estimates as well as attain the posterior Cramer-Rao lower bound (CRLB).
The figure shows a simulation run of PROP, where the true and estimated reflecting surfaces, propagation paths and agent positions are shown for time n = 18 (a,b) and n = 307 (c,d), respectively. The line representation of estimated surfaces are computed using the MMSE estimates of the detected surface features. Estimated propagation paths are obtained by connecting the MMSE estimates of the agent position, interaction points on the estimated surfaces and PAs, and compared against the true visible paths at each step. The color of each estimated path represents its SNR estimates, i.e., the square of norm amplitude MMSE estimates.
If you are interested have a look here.
Browse the Results of the Month archive.