A Blind Source Separation Approach Based on Normalized Convex Perimeter

This paper addresses the problem of blind source separation for both independent and dependent sources. Signals in wireless communication system usually own a bounded nature, in view of this observation, a method based on bounded component analysis (BCA) for communication signals separation is proposed. The normalized convex perimeter is adopted as the contrast function and the algorithm is further optimized by a gradient decent algorithm. Experimental results show that the proposed algorithm outperforms the existent BCA algorithms and obtains superior performance over the state of art independent component analysis (ICA)-based algorithms for a small number of samples in high SNR scenarios.