Mitigating the impact of non-ideal PRBS on photonics-assisted compressive sensing systems through complementary PRBS and parallel links

Photonic random demodulator (RD) is a widely studied scheme to realize photonics-assisted compressive sensing (CS). Since CS via an RD requires a pseudo-random binary sequence (PRBS) bit rate that is at least twice the signal's Nyquist rate, the generation of the ideal PRBS in practical systems requires a very large signal bandwidth and is difficult to achieve, leading to poor mixing results from the RD and limited performance of the photonics-assisted CS system. To solve this problem, in this work, we propose a method to compensate for the poor mixing results by using a complementary PRBS and a parallel optical path, thus achieving better CS results in the case of limited PRBS bandwidth. The feasibility of the proposed method is verified by simulation. When only a 4-Gbps PRBS with a bandwidth limited to 0 to 6 GHz is used and the signal to be sampled is a multi-tone signal lower than 1.6 GHz, it is difficult to reconstruct the multi-tone signal or the quality of the reconstructed signal is poor when the starting point is randomly selected for downsampling due to the non-ideal PRBS. When the complementary PRBS is also limited to 0 to 6 GHz, the quality of the reconstructed signal is also improved in most cases. The method proposed in this work provides a good solution for the realization of CS using photonics-assisted RD in real systems, which can reduce the influence of non-ideal PRBS on signal recovery without expanding the bandwidth of the used PRBS.