Reducing the Computational Complexity of Reconstruction in Compressed Sensing Nonuniform Sampling

This paper proposes a method that reduces the computational complexity of signal reconstruction in single-channel nonuniform sampling while acquiring frequency sparse multi-band signals. Generally, this compressed sensing based signal acquisition allows a decrease in the sampling rate of frequency sparse signals, but requires computationally expensive reconstruction algorithms. This can be an obstacle for real-time applications. The reduction of complexity is achieved by applying a multi-coset sampling procedure. This proposed method reduces the size of the dictionary matrix, the size of the measurement matrix and the number of iterations of the reconstruction algorithm in comparison to the direct single-channel approach. We consider an orthogonal matching pursuit reconstruction algorithm for single-channel sampling and its modification for multi-coset sampling. Theoretical as well as numerical analyses demonstrate order of magnitude reduction in execution time for typical problem sizes without degradation of the signal reconstruction quality.