Greedy tree search for Internet of Things signal detection

In this paper, we propose a greedy tree search algorithm for Internet of Things (IoT) signal detection. The proposed method, referred to as matching pursuit with integer tree search (MP-ITS), recovers integer sparse vector (sparse vector whose nonzero elements are chosen from a set of finite alphabets) using the tree search. In order to control the computational burden yet maintains the effectiveness of the tree search, MP-ITS employs two strategies, viz., pre-screening to put a limitation on columns of the channel matrix and tree pruning to eliminate unpromising candidates from the tree. We show from the restricted isometry property (RIP) analysis and empirical simulations on realistic IoT scenarios that the proposed method is effective in recovering the sparse vector with integer constraint.