Signal estimation for interference cancellation and decision feedback equalization

A signal estimation device for antipodal signals is derived which is optimal in the sense of minimizing the mean square estimation error. It is applied to estimate the interfering signal components in a received signal in order to cancel them to improve the detection performance. It is shown, that especially for bad signal-to-noise ratios, the optimum device, which delivers a soft value for feedback, improves the quality of interference elimination compared to the standard two point characteristic (sign-function=hard feedback). The performance enhancement using the optimum device is investigated in two possible applications: interference cancellation for CDMA signals and decision feedback equalization for signals in frequency selective channels. A considerable gain is achieved in the first case whereas in the latter case it is only marginal. This fact is explained theoretically.