Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Time-Frequency Analysis of Variable-Length WiFi CSI Signals for Person Re-Identification

Chen Mao, Chong Tan, Jingqi Hu, Min Zheng

TL;DR

This work tackles privacy-friendly person re-identification by leveraging WiFi CSI signals. It presents a two-stream transformer network that separately analyzes time-domain amplitude and frequency-domain phase from variable-length CSI data, with continuous lateral connections to fuse multi-level time-frequency features. LMCL and SoftTriple losses are employed to enforce discriminative representations and robust metric learning. On the ViFi-Indoors dataset, the method delivers high accuracy (e.g., mAP ≈ 93.68%, Rank-1 ≈ 98.13%) and demonstrates robustness through ablations on data augmentation and fusion strategies, highlighting practical potential for infrastructure-based ReID without visual data.

Abstract

Person re-identification (ReID), as a crucial technology in the field of security, plays an important role in security detection and people counting. Current security and monitoring systems largely rely on visual information, which may infringe on personal privacy and be susceptible to interference from pedestrian appearances and clothing in certain scenarios. Meanwhile, the widespread use of routers offers new possibilities for ReID. This letter introduces a method using WiFi Channel State Information (CSI), leveraging the multipath propagation characteristics of WiFi signals as a basis for distinguishing different pedestrian features. We propose a two-stream network structure capable of processing variable-length data, which analyzes the amplitude in the time domain and the phase in the frequency domain of WiFi signals, fuses time-frequency information through continuous lateral connections, and employs advanced objective functions for representation and metric learning. Tested on a dataset collected in the real world, our method achieves 93.68% mAP and 98.13% Rank-1.

Time-Frequency Analysis of Variable-Length WiFi CSI Signals for Person Re-Identification

TL;DR

This work tackles privacy-friendly person re-identification by leveraging WiFi CSI signals. It presents a two-stream transformer network that separately analyzes time-domain amplitude and frequency-domain phase from variable-length CSI data, with continuous lateral connections to fuse multi-level time-frequency features. LMCL and SoftTriple losses are employed to enforce discriminative representations and robust metric learning. On the ViFi-Indoors dataset, the method delivers high accuracy (e.g., mAP ≈ 93.68%, Rank-1 ≈ 98.13%) and demonstrates robustness through ablations on data augmentation and fusion strategies, highlighting practical potential for infrastructure-based ReID without visual data.

Abstract

Person re-identification (ReID), as a crucial technology in the field of security, plays an important role in security detection and people counting. Current security and monitoring systems largely rely on visual information, which may infringe on personal privacy and be susceptible to interference from pedestrian appearances and clothing in certain scenarios. Meanwhile, the widespread use of routers offers new possibilities for ReID. This letter introduces a method using WiFi Channel State Information (CSI), leveraging the multipath propagation characteristics of WiFi signals as a basis for distinguishing different pedestrian features. We propose a two-stream network structure capable of processing variable-length data, which analyzes the amplitude in the time domain and the phase in the frequency domain of WiFi signals, fuses time-frequency information through continuous lateral connections, and employs advanced objective functions for representation and metric learning. Tested on a dataset collected in the real world, our method achieves 93.68% mAP and 98.13% Rank-1.
Paper Structure (8 sections, 6 equations, 5 figures, 3 tables)

This paper contains 8 sections, 6 equations, 5 figures, 3 tables.

Table of Contents

  1. Introduction
  2. Proposed Method
  3. Data Preprocessing
  4. Two-Stream Network Architecture
  5. Time-Frequency Feature Fusion
  6. Representation and Metric Learning
  7. Experiments
  8. Conclusion

Figures (5)

  • Figure 1: Schematic diagram of WiFi signal propagation to the human body.
  • Figure 2: Original, unwrapped and linear transformed phase visualization.
  • Figure 3: Architecture of the two-stream time-frequency person ReID network based on WiFi CSI signals.
  • Figure 4: After normalizing and padding the length of the signal data, the base embeddings and position embeddings are merged to form a complete input.
  • Figure 5: Visualizations of pedestrian feature clustering in two-dimensional space and ROC curve.