Underdetermined Blind Recovery of Communication Signals Based on Minimum Euclidean Distance in Time-Frequency Domain

Abstract

To recover the source signals in underdetermined case is a challenging problem, especially when the source signals are non-disjoint in Time-Frequency (TF) domain. The conventional algorithms such as subspace-based complete the blind recovery utilizing the sparsity of the original signals with the assumption that the number of active sources at any TF point is strictly less than that of sensors. Moreover, the processed signals are speech or image signals usually because these signals are sparse in TF domain. But for communication signals, the overlapping amount become more serious in TF domain, so that the performances of conventional algorithms are weaken. Considering the continuity of communication signals in TF domain, this paper proposes a new method to recover communication signals which relaxes the sparsity condition of sources in TF domain. The method allows that the number of active sources at any TF point simultaneously equals to the number of sensors. We can identify the active sources and estimate their corresponding TF values at any TF point by calculating the Euclidean distances between the detected point and all single source points. The computer simulations show that the proposed estimation algorithm is more efficient than the previous algorithms.