The Restricted Isometry Property for Banded Random Matrices

Abstract

In this paper, we investigate the problem of determining the conditions under which the restricted isometry property (RIP) is satisfied for a particular type of matrix referred to in here as a banded random matrix (BRM). Such matrices have been recognized as suitable models for a number of compressive-sensing based sampling architectures, including the interleaved random demodulator, the random demodulator, the parallel non-interleaved random demodulator, the random sampler, and the periodic nonuniform sampler. It is thus important to establish the conditions under which the BRM satisfies the RIP; to our knowledge, this question has not been theoretically addressed in the literature. If the resulting conditions are satisfied, full signal recovery using a convex optimization algorithm is guaranteed. The specific objective of this research is to determine the conditions under which the RIP is satisfied for two possible sampling matricies: a BRM and a BRM multiplied by the discrete Fourier matrix.