SigChord: Sniffing Wide Non-sparse Multiband Signals for Terrestrial and Non-terrestrial Wireless Networks

TL;DR

SigChord tackles the challenge of sniffing wide, non-sparse wireless spectra under sub-Nyquist constraints by marrying a Transformer-based spectrum sensing module with a rule-based least-squares recovery, followed by cascaded Transformer-based protocol identification and header decoding. By predicting spectrum occupancy and recovering signals below the Landau-rate bound, SigChord breaks traditional sampling-rate limits and enables detailed physical-layer inspection of concurrent terrestrial and non-terrestrial signals. Across synthetic and over-the-air datasets, it achieves over 99% accuracy in detection and decoding while reducing the sampling rate relative to state-of-the-art methods, and it runs in real time on consumer-grade GPUs. This holistic sniffing pipeline provides comprehensive, open, and scalable insights into modern wireless environments, with practical implications for network optimization, coexistence, and security research.

Abstract

While unencrypted information inspection in physical layer (e.g., open headers) can provide deep insights for optimizing wireless networks, the state-of-the-art (SOTA) methods heavily depend on full sampling rate (a.k.a Nyquist rate), and high-cost radios, due to terrestrial and non-terrestrial networks densely occupying multiple bands across large bandwidth (e.g., from 4G/5G at 0.4-7 GHz to LEO satellite at 4-40 GHz). To this end, we present SigChord, an efficient physical layer inspection system built on low-cost and sub-Nyquist sampling radios. We first design a deep and rule-based interleaving algorithm based on Transformer network to perform spectrum sensing and signal recovery under sub-Nyquist sampling rate, and second, cascade protocol identifier and decoder based on Transformer neural networks to help physical layer packets analysis. We implement SigChord using software-defined radio platforms, and extensively evaluate it on over-the-air terrestrial and non-terrestrial wireless signals. The experiments demonstrate that SigChord delivers over 99% accuracy in detecting and decoding, while still decreasing 34% sampling rate, compared with the SOTA approaches.

Figures (19)

• Figure 1: The SigChord wide multiband sniffing system. Unlike sequential scanning methods shi2015beyondguddeti2019sweepsensesubbaraman2023crescendowireshark that capture only one signal at a time, or sub-Nyquist methods mishali2009blindhassanieh2014ghzqin2018sparsesong2022approaching limited to sparse spectra, SigChord detects and decodes concurrent, non-sparse wide multiband signals at low sampling rates.
• Figure 2: Nyquist rate, Landau rate landau1967necessary and blind sub-Nyquist rate mishali2009blind for multiband signals. The Nyquist rate equals to the bandwidth of the whole spectrum. The Landau rate equals to the sum of the bandwidth of each signal. The blind sub-Nyquist rate in mishali2009blind is twice the Landau rate.
• Figure 3: CS recovered tropp2005simultaneous spectrum under different sampling rates with SNR of 10 dB. (a) Original; (b) Recovered at 146% of the blind sub-Nyquist limit; (c) Recovered at 83% of the blind sub-Nyquist limit.
• Figure 4: Signal processing pipeline of SigChord
• Figure 5: The embedding process of SigChord.