School of Information Science and Engineering，Yunnan University，Kunming 650500，China
Abstract：At present, the compressed sampling system mainly includes three sampling system: random demodulator，modulated wideband converter and multi-coset sampling, among them, the system framework of multi-coset sampling is simple and has a high reconstruction success rate. But, the multi-coset sampling faces two difficulties in hardware implementation, one is that there is an upper limit on the input analog bandwidth of the low-speed analog-to-digital converter(ADC), and the other is it′s difficult to achieve precise delays on each channel. This paper draws on the ideal of modulated wideband converter which the input signal is pre-modulated before sampling, proposes the pre-modulated multi-coset sampling system which combines pre-modulation with multi-coset sampling. Before sampling, the generate periodic single-pulse sequences is generated by a random sequence generator in each channel of the sampling system, and the input signal is spectrally modulated to reduce the analog bandwidth of the signal before sampling. Then, the modulated signal is sampled by the low-speed ADC, and each channel of low-speed ADC and random sequence generator share a same control clock. Finally, the multiple signal classification algorithm(MUSIC) is used to reconstruct the input signal without distortion. The simulation results show that the mean square error of the reconstructed signal is at the order of 10-5 when the input signal is compressed and sampled at 15% Nyquist sampling rate. This verifies the feasibility of the sampling system in this paper, and the sampling system effectively overcome the two problems of the multi-coset sampling system.
Key words :compressed sensing；multi-coset sampling；pre-modulated；MUSIC algorithm；modulated wideband converter
