A grid-free frequency estimation method using multi-channel co-prime sampling scheme

This paper proposes an effective method for frequency spectrum estimation using multi-channel co-prime sampling. Previous work has adopted a multi-channel sampling system to sense sparse signals, but its model suffers from the grid mismatch problem and the average sampling rate is high due to the redundancy among different channels. In this paper, we design a multi-channel co-prime sampling scheme to further reduce the average sampling rate. Then we incorporate the grid-free method called super-resolution for continuous support recovery which is not influenced by the discretized grids. Meanwhile, we also give a new support set reduction criterion which further decreases the cardinality of candidate frequency support set and the optimum choice of co-prime sampling rate. Numerical results demonstrate the correctness of our proposed method.