Research On Photonic-assisted Compressive Sampling Technique

Compressive sampling (CS) is well known for sampling sparse signals at sub-Nyquistrate, and it has been applied extensively in radar signal sensing. To increase thefrequency and bandwidth of CS system further, broadband and low-jitter photonictechniques are introduced into CS. In current, with the help of photonic links, these twoparameters have been improved a lot. However, there are many important issues to besolved, in aspect of system implementation and performance optimization. This paperfocuses on four key issues including the implementation of multi-channel photonic CSsystem, optimization of signal recovery rate, bandwidth extension and generation ofmixing signal between microwave and±1level pseudo-random bit sequence (PRBS)signal.To overcome the difficulty of multi-channel photonic CS system implementation, anovel simplified CS scheme is proposed based on multi-wavelength time-delayed PRBSsignals, where multiple PRBS signals share one PRBS generator and modulator.Meanwhile, the noise and distortion of mixed signal induced by photonics in system areanalyzed. Based on these preparations, an eight-channel photonic CS system with5-GHz bandwidth is set up and demonstrated firstly.To increase the recovery rate of random sparse spectrum, the restricted isometryproperty of sensing matrix used in CS system is optimized by generating multi-channelPRBS signals with nonlinear time delays. Through large amount of numericalsimulation, it is found that the sensing matrix, constructed by Fourier coefficients ofnonlinearly time-delayed multi-channel PRBS signals, could satisfy the RIP with just30%more measurements than that required by the sensing matrix corresponding tomulti-different PRBSs. Due to this improvement, this simplified scheme could supporthigh recovery possibility of random sparse signals with narrow-bandwidth samples. Inexperiment, with four-wavelength nonlinearly time-delayed PRBS signals, the sparsespectrums in the0.2-5-GHz range are recovered faithfully from four-channelcompressed signals with only120-MHz bandwidth. Meanwhile, multi-channel PRBSsignals with tunable time delays are also proposed to apply for guaranteeing effectiverecovery of random multi-band signals with prior knowledge of band number from compressed signal with information-rate bandwidth. In simulation, random two-bandsignals are recovered from samples with information-rate bandwidth.Another primary contribution of this paper is to apply time-interleaved PRBSpulses to extend CS bandwidth without limit of the repetition rate of optical PRBS pulse.The feasibility of this proposal is verified by the simulation on the RIP of sensing matrix.With this theoretical support, a four-channel CS with20GHz bandwidth isdemonstrated with10.16-Gbps PRBS pulses, where signals occupying different bandsare identified simultaneously.To obtain the mixed signal between±1level PRBS signal and microwave, ascheme based on phase modulation and coherent detection is presented. The errorsources of mixed signal resulted from photonical apparatuses are also analyzed. Basedon this knowledge, an experimental photonic CS is set up, where multi-tone signaldistributing in0-5-GHz range is recovered faithfully from compressed spectrum withonly200-MHz bandwidth.