FRI Sampling of Parametric Signals with Non-ideal Sinc Kernel

Recent developed finite rate of innovation (FRI) theory provides an efficient way for sub-Nyquist sampling of wideband parametric signals in the fields of radar and communication. However, the existing hardware schemes of FRI sampling systems have not considered the non-ideal effects of physical components, which lead to low accuracy in the signal reconstruction process. In this paper, we propose a FRI sampling system for parametric signals with non-ideal sinc kernel, which significantly improves the reconstruction performance under non-ideal hardware environment. The proposed system consists of two parallel sampling channels. The first channel samples the parametric signal at a low speed after filtered with a LPF, which is used to obtain a part of Fourier coefficients. The other channel obtains a part of Fourier coefficients of its basis signal in the same way. To eliminate the non-ideal effects of physical components, we propose a parameters joint estimation algorithm by using the obtained Fourier coefficients. We also provide a hardware platform design scheme for implementing our system. According to simulation and hardware experiment results, the proposed system can eliminate non-ideal effects in the hardware process, and greatly improve the signal reconstruction accuracy.