Design and Development of Multirate Approach for SPARSE Signal Estimation
This paper proposed a novel compressive sensing framework for sparse signals. In this approach, sparse signals sampled through multirate multichannel system. In most of the signal recovery methods, it has been assumed that there was no noise present in the system. However, in realistic scenario, the noise will be added in the signal-acquisition devices. In this article, theprinciples of filter bank are appliedfor CS acquisition of speech signal. The sensing method comprises of a filter bank of finite impulse response (FIR) filters followed bydecimation. The filter bank decimation factor can beadjusted to sample at the Nyquist rate (maximally decimated)or to sample below the Nyquist rate (over-decimated). A more realistic model in which the observations are contaminated by an additive noise is considered. The performance of the OMP algorithm and a filter bank based new M-estimator technique are compared. It is verified that new M-estimator technique offers a significant performance when the measurements are corrupted by noise.