Multi-BS PHD-SLAM: A Computationally Efficient EKF-LoS/NLoS Fusion Framework for RF Sensing

View at publisher Summarise with VIRAT
Authors: Tao Du, Jie Yang, Hang Que, Chao-Kai Wen, Shuqiang Xia, Christos Masouros, Shi Jin
Cited by: 0
References: 42

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

Integrated Sensing and Communication (ISAC) has the potential to enhance both energy and spectral efficiency in modern communication systems. Although Probability Hypothesis Density (PHD)-based Simultaneous Localization and Mapping (SLAM) is a key algorithm for positioning and environmental mapping in ISAC, the advantages of multi-base-station (multi-BS) fusion remain underexplored, despite the considerable attention given to multi-sensor and multi-user data fusion in existing research. This paper leverages the distinct roles of Line-of-Sight (LoS) and Non-Line-of-Sight (NLoS) channel parameters, employing LoS for agent localization and NLoS for environment mapping. An Extended Kalman Filter (EKF) framework is proposed to fuse LoS path angle parameters for localization, for which the corresponding Cramér-Rao Lower Bound (CRLB) is derived. To facilitate landmark mapping, a virtual reference point (VRP) is introduced to model reflecting surfaces consistently across base stations (BSs), replacing the conventional approach of using multiple virtual anchors for multiple BSs. Furthermore, map fusion algorithms are developed to address the challenges of merging PHD-SLAM maps with varying observation quality and overlapping fields of view. To reduce the computational complexity of particle-based PHD-SLAM, agent location estimates derived from EKF fusion are used as priors, significantly improving particle efficiency and enabling the unified exploitation of LoS and NLoS data for comprehensive situational awareness. Simulation and experimental results confirm that the proposed EKF-based LoS fusion strategy significantly improves sensing performance while maintaining low computational overhead.

Fields of Study & Subjects

Indoor and Outdoor Localization Technologies Robotics and Sensor-Based Localization Target Tracking and Data Fusion in Sensor Networks

Publication details

Journal / source
IEEE Transactions on Wireless Communications
Volume
25
Pages
14152–14165
Publication date
1 January 2026
Language
EN