Photonic Compressive Sensing System Based on 1-Bit Quantization for Broadband Signal Sampling

Photonic compressive sensing (PCS), a paradigm for broadband sparse radio frequency (RF) signal acquisition, is recognized for its ability to significantly reduce sampling rate. However, conventional PCS systems face several challenges, including high storage requirements, limited noise resistance, and increased energy consumption during sampling. Additionally, they often struggle to accurately reconstruct frequency non-sparse signals and typically fail to resolve the time-frequency characteristics of signals with long temporal durations. In this study, we propose and experimentally validate a novel 1-bit quantization PCS (1-bit PCS) framework. Our system employs dictionary learning and local compressive sensing algorithm, enabling the effective recovery of various signal types, such as multi-tone, step-frequency, and linear frequency modulation (LFM) signals across a wide frequency range of 0 to 5 GHz. The sampling rate achieved is 1 GSa/s, corresponding to a compression ratio of 10. With this system, not only can th e time-domain waveforms and spectrum of signals be accurately recovered, but the time-frequency characteristics are also captured with precision. This makes the system highly promising for applications in resource-limited scenarios.