Application Of The Score Algorithm And Score Extensions To Sorting In The Rank-L Spectral Self-coherence Environment

The Self Coherence REstoral (SCORE) algorithm has been shown to blindly extract a single signal of interest with spectra correlation at a known value of frequency separation (e.g., baud rate) from an arbitrary number of interferers without correlation at that frequency separation. This paper investigates the ability of the SCORE approach to extract cyclostationary signals from a rank L spectral self-coherence environment where L signals exhibit spectral correlation at the same frequency separation. It is shown analytically and by computer simulation that the SCORE algorithm can separate multiple signals with spectral self-coherence at the same value of frequency separation, provided that those signals have different self-coherence strengths. An extension of the SCORE algorithm is also developed that can separate sig- nals if their complex-valued spectral self-coherence functions are different, e.g., in a communications net where the signals have dif- ferent timing phases. It is shown that the SINR of the separated signals approaches the maximum attainable SINR.