Measurement Matrix for Energy Efficient Compressive Sensing

Compressive Sensing (CS) is an emerging strategy for non-adaptive sub-Nyquist sampling of sparse signals, where the sampling is achieved through a set of linear random projections of the original signals, and the signal recovery can be realized via convex optimization method or greedy algorithms. CS involves relatively simple computation of on-chip hardware and leaves complex signal recovery of computational demanding for off-chip processing. The implementation of measurement matrix is critical to the design of energy efficient CS. According to different applications, various efficient hardware realizations of measure matrix are investigated, including pseudorandom bit sequence, no-uniform sampling, sparse binary matrix, and stochastic oscillators. other potential hardware efficient approaches to improve CS energy efficiency are also proposed, such as circulant matrix and adaptive CS.