Signal recovery from noisy samples via multiple observations and thresholding

The problem of sampling and recovering bandlimited signals in the presence of noise is studied. A new technique based on acquisition of multiple sets of sampled values at the Nyquist sampling rate and thresholding-based reconstruction is proposed. The exact formula and upper bound for mean integrated squared error of the proposed scheme is established. The obtained results show that the proposed technique gives better reconstruction accuracy than the common reconstruction method using oversampling and post-filtering.