Coherent Direct Multipath SLAM

TL;DR

This paper introduces a scalable Bayesian direct MP-SLAM method for coherent data fusion in D-MIMO/XL-MIMO systems, enabling high-accuracy localization and environment mapping directly from raw RF signals.

Contribution

It proposes a novel phase-preserving likelihood function combined with a surface feature vector model for coherent, distributed RF signal processing in MIMO systems.

Findings

Performance surpasses existing noncoherent methods.

Approaches the posterior CRLB in simulations.

Supports near real-time processing with GPU acceleration.

Abstract

Challenging indoor and urban environments with severe multipath propagation and obstructed LoS (OLoS) degrade classical radio frequency (RF) positioning. Multipath-based simultaneous localization and mapping (MP-SLAM) is a promising remedy, building and exploiting a map of the propagation environment to enhance the robustness. Emerging distributed multiple-input multiple-output (D-MIMO)/extremely large-scale MIMO (XL-MIMO) infrastructures, with single XL antenna arrays or distributed subarrays, offer large spatial apertures and enable high-resolution sensing, in particular when phase coherence is maintained across base stations (BSs), subarrays, or distributed arrays. In this work, we propose a scalable Bayesian direct MP-SLAM method for coherent data fusion in D-MIMO/XL-MIMO systems that jointly infers the environment while performing robust, high-accuracy localization directly from raw RF signals. The key idea is a phase-preserving nonzero-mean Type-II likelihood function in which a complex mean is shared across BSs or subarrays and enables coherent fusion, while the variance captures noncoherent signal power. The likelihood function is combined with a surface feature vector (SFV)-based model that enables map feature fusion across the distributed infrastructure and supports near-field propagation and visibility effects. A GPU-parallel implementation enables highly scalable processing across a distributed infrastructure and particles, possibly allowing real-time calculations for large antenna arrays. Simulation results demonstrate performance gains over existing noncoherent methods and approach the corresponding posterior CRLB (PCRLB), highlighting the potential of coherent distributed arrays for high-resolution sensing and localization.