BLIND SOURCE SEPARATION FOR TWO-DIMENSION SPREAD SPECTRUM SYSTEM BASED ON TRILINEAR DECOMPOSITION

This paper links the two-dimension-spread-spectrum-system source separation problem to the trilinear model, which is an analysis tool rooted in psychometrics and chemometrics. Exploiting this link, it derives a blind source separation algorithm. The proposed algorithm capitalizes on time-domain spread, frequency-domain spread and temporal diversity-combining. The simulation results reveal that the performance of the blind source separation algorithm for two-dimension spread spectrum system is very close to nonblind minimum mean-squared error method, and this algorithm works well for small sample size. The blind source separation algorithm does not require channel fading information and spread codes, so it has blind and robust characteristics.