Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

A Novel Cross-band CSI Prediction Scheme for Multi-band Fingerprint based Localization

Yuan Ruihao, Huang Kaixuan, Zhang Shunqing

TL;DR

This work tackles the challenge of high-mobility localization using multi-band fingerprints by formulating it as a mobility-aware channel prediction problem. It introduces a two-stage approach: (i) MUSIC-assisted, SAGE-based estimation of per-path mobility parameters to capture dynamic effects, and (ii) a VAE-based cross-band channel reconstruction that transfers static CSI across bands, augmented by a mobility-removal block. The method demonstrates superior parameter estimation and robust cross-band channel prediction, yielding localization errors that closely approach the multi-band performance bound and outperform single-band baselines. The combination of mathematical grounding and data-driven reconstruction provides a practical means to exploit fingerprint diversity while managing sampling complexity in ISAC applications.

Abstract

Because of the advantages of computation complexity compared with traditional localization algorithms, fingerprint based localization is getting increasing demand. Expanding the fingerprint database from the frequency domain by channel reconstruction can improve localization accuracy. However, in a mobility environment, the channel reconstruction accuracy is limited by the time-varying parameters. In this paper, we proposed a system to extract the time-varying parameters based on space-alternating generalized expectation maximization (SAGE) algorithm, then used variational auto-encoder (VAE) to reconstruct the channel state information on another channel. The proposed scheme is tested on the data generated by the deep-MIMO channel model. Mathematical analysis for the viability of our system is also shown in this paper.

A Novel Cross-band CSI Prediction Scheme for Multi-band Fingerprint based Localization

TL;DR

This work tackles the challenge of high-mobility localization using multi-band fingerprints by formulating it as a mobility-aware channel prediction problem. It introduces a two-stage approach: (i) MUSIC-assisted, SAGE-based estimation of per-path mobility parameters to capture dynamic effects, and (ii) a VAE-based cross-band channel reconstruction that transfers static CSI across bands, augmented by a mobility-removal block. The method demonstrates superior parameter estimation and robust cross-band channel prediction, yielding localization errors that closely approach the multi-band performance bound and outperform single-band baselines. The combination of mathematical grounding and data-driven reconstruction provides a practical means to exploit fingerprint diversity while managing sampling complexity in ISAC applications.

Abstract

Because of the advantages of computation complexity compared with traditional localization algorithms, fingerprint based localization is getting increasing demand. Expanding the fingerprint database from the frequency domain by channel reconstruction can improve localization accuracy. However, in a mobility environment, the channel reconstruction accuracy is limited by the time-varying parameters. In this paper, we proposed a system to extract the time-varying parameters based on space-alternating generalized expectation maximization (SAGE) algorithm, then used variational auto-encoder (VAE) to reconstruct the channel state information on another channel. The proposed scheme is tested on the data generated by the deep-MIMO channel model. Mathematical analysis for the viability of our system is also shown in this paper.
Paper Structure (14 sections, 1 theorem, 12 equations, 4 figures, 2 tables)

This paper contains 14 sections, 1 theorem, 12 equations, 4 figures, 2 tables.

Table of Contents

  1. Introduction
  2. System Model and Preliminary
  3. System Model
  4. Problem Formulation
  5. Solution Based on SAGE and VAE
  6. Modified SAGE Algorithm for Parameters Estimation
  7. Deep Learning Based Channel Reconstruction
  8. Experiment Results
  9. Parameters Estimation
  10. Channel Reconstruction
  11. Localization Results
  12. Acknowledge
  13. Conclusion
  14. Proof of Lemma \ref{['lem:equ']}

Key Result

Lemma 1

Problem 1 and Problem 3 are linear connected and when the errors of the two problems are minimized, the error of problem 2 is also minimized.

Figures (4)

  • Figure 1: The system model.
  • Figure 2: The whole procedure of the system.
  • Figure 3: CDF of CCNE in channel reconstruction.
  • Figure 4: CDF of localization error.

Theorems & Definitions (1)

  • Lemma 1