RayLoc: Wireless Indoor Localization via Fully Differentiable Ray-tracing

TL;DR

RayLoc tackles indoor localization by recasting it as an inverse problem over a fully differentiable ray-tracing simulator. By jointly optimizing trainable scene parameters and target location, it reconstructs a high-fidelity physical environment to generate CSI that matches measurements, using gradient-based methods enhanced by Gaussian smoothing to avoid local minima. The approach unifies device-free and device-based localization and demonstrates superior accuracy and generalization across diverse real-world environments, even under NLoS conditions. Practically, RayLoc enables precise positioning with moderate infrastructure and offers a path toward scalable, scene-aware RF sensing in complex indoor spaces.

Abstract

Wireless indoor localization has been a pivotal area of research over the last two decades, becoming a cornerstone for numerous sensing applications. However, conventional wireless localization methods rely on channel state information to perform blind modelling and estimation of a limited set of localization parameters. This oversimplification neglects many sensing scene details, resulting in suboptimal localization accuracy. To address this limitation, this paper presents a novel approach to wireless indoor localization by reformulating it as an inverse problem of wireless ray-tracing, inferring scene parameters that generates the measured CSI. At the core of our solution is a fully differentiable ray-tracing simulator that enables backpropagation to comprehensive parameters of the sensing scene, allowing for precise localization. To establish a robust localization context, RayLoc constructs a high-fidelity sensing scene by refining coarse-grained background model. Furthermore, RayLoc overcomes the challenges of sparse gradient and local minima by convolving the signal generation process with a Gaussian kernel. Extensive experiments showcase that RayLoc outperforms traditional localization baselines and is able to generalize to different sensing environments.

Figures (16)

• Figure 1: Unlike traditional wireless localization performing blind estimation of a few parameters based on CSI, RayLoc infers the full-set scene parameters by the inverse process of differentiable wireless ray-tracing.
• Figure 2: Discrepancies between pre-change and post-change power delay profiles arise from inaccuracies in locations and EM properties.
• Figure 3: Optimization-unfriendly loss landscape of RT.
• Figure 4: The workflow of RayLoc.
• Figure 5: Illustration of the SBR process.