Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

MARBLE-Net: Learning to Localize in Multipath Environment with Adaptive Rainbow Beams

Qiushi Liang, Yeyue Cai, Jianhua Mo, Meixia Tao

TL;DR

MARBLE-Net tackles high-precision localization in multipath ISAC by jointly optimizing a frequency-dependent rainbow beamformer and a CNN-based locator in an end-to-end framework. The approach treats phase shifters $\phi$ and true-time-delays $\tau$ as trainable parameters, enabling adaptive beam squint to exploit environment-specific multipath fingerprints. Through a three-stage training strategy and ray-tracing–generated datasets, MARBLE-Net achieves substantial localization gains over fixed-beam baselines and a k-NN codebook, including sub-meter performance in structured multipath scenes where direct LOS is not dominant. The work demonstrates that co-designing physical-layer sensing beams with neural processing can turn multipath from a hindrance into a localization resource, with real-time inference and scalable complexity, and it points to extensions to 3D, multi-target, and dynamic environments.

Abstract

Integrated sensing and communication (ISAC) systems demand precise and efficient target localization, a task challenged by rich multipath propagation in complex wireless environments. This paper introduces MARBLE-Net (Multipath-Aware Rainbow Beam Learning Network), a deep learning framework that jointly optimizes the analog beamforming parameters of a frequency-dependent rainbow beam and a neural localization network for high-accuracy position estimation. By treating the phase-shifter (PS) and true-time-delay (TTD) parameters as learnable weights, the system adaptively refines its sensing beam to exploit environment-specific multipath characteristics. A structured multi-stage training strategy is proposed to ensure stable convergence and effective end-to-end optimization. Simulation results show that MARBLE-Net outperforms both a fixed-beam deep learning baseline (RaiNet) and a traditional k-nearest neighbors (k-NN) method, reducing localization error by more than 50\% in a multipath-rich scene. Moreover, the results reveal a nuanced interaction with multipath propagation: while confined uni-directional multipath degrades accuracy, structured and directional multipath can be effectively exploited to achieve performance surpassing even line-of-sight (LoS) conditions.

MARBLE-Net: Learning to Localize in Multipath Environment with Adaptive Rainbow Beams

TL;DR

MARBLE-Net tackles high-precision localization in multipath ISAC by jointly optimizing a frequency-dependent rainbow beamformer and a CNN-based locator in an end-to-end framework. The approach treats phase shifters and true-time-delays as trainable parameters, enabling adaptive beam squint to exploit environment-specific multipath fingerprints. Through a three-stage training strategy and ray-tracing–generated datasets, MARBLE-Net achieves substantial localization gains over fixed-beam baselines and a k-NN codebook, including sub-meter performance in structured multipath scenes where direct LOS is not dominant. The work demonstrates that co-designing physical-layer sensing beams with neural processing can turn multipath from a hindrance into a localization resource, with real-time inference and scalable complexity, and it points to extensions to 3D, multi-target, and dynamic environments.

Abstract

Integrated sensing and communication (ISAC) systems demand precise and efficient target localization, a task challenged by rich multipath propagation in complex wireless environments. This paper introduces MARBLE-Net (Multipath-Aware Rainbow Beam Learning Network), a deep learning framework that jointly optimizes the analog beamforming parameters of a frequency-dependent rainbow beam and a neural localization network for high-accuracy position estimation. By treating the phase-shifter (PS) and true-time-delay (TTD) parameters as learnable weights, the system adaptively refines its sensing beam to exploit environment-specific multipath characteristics. A structured multi-stage training strategy is proposed to ensure stable convergence and effective end-to-end optimization. Simulation results show that MARBLE-Net outperforms both a fixed-beam deep learning baseline (RaiNet) and a traditional k-nearest neighbors (k-NN) method, reducing localization error by more than 50\% in a multipath-rich scene. Moreover, the results reveal a nuanced interaction with multipath propagation: while confined uni-directional multipath degrades accuracy, structured and directional multipath can be effectively exploited to achieve performance surpassing even line-of-sight (LoS) conditions.

Paper Structure

This paper contains 18 sections, 2 equations, 6 figures, 3 tables.

Table of Contents

  1. Introduction
  2. System and Channel Model
  3. Rainbow Beamforming
  4. High-Fidelity Ray-Tracing Channel Generation
  5. Proposed Localization Framework
  6. Learnable Rainbow Beamformer ($\mathcal{M}_1$)
  7. Prediction Network ($\mathcal{M}_2$)
  8. Multi-Stage Training Strategy
  9. Simulation Results
  10. Simulation Setup
  11. Baseline
  12. k-Nearest Neighbors (k-NN)
  13. RaiNet
  14. Performance Comparison and Ablation Study
  15. Qualitative and Error Distribution Analysis
  16. ...and 3 more sections

Figures (6)

  • Figure 1: System architecture for uplink UAV localization with a phase-time array at the BS.
  • Figure 2: Three multipath scenes modeled by Blender.
  • Figure 3: Data flow of the MARBLE-Net architecture.
  • Figure 4: Spatial distribution of the localization error of the MARBLE-Net (adaptive) at $P_t=23dBm$ across the four scenes.
  • Figure 5: Localization error CDF of MARBLE-Net at $P_t = 23$ dBm.
  • ...and 1 more figures