Estimating noise statistics in adaptive semi-blind equalization

In our previous work we have developed semi-blind equalization algorithms for time varying channels. These methods combine decision feedback equalization (DFE) with Kalman filtering for channel tracking. Methods using Kalman filtering commonly assume that process and measurement noise statistics are known. This is not the case in practice. In this paper we propose a method for estimating these statistics using low complexity batch processing. Reliability of the estimates is verified by testing the whiteness of the innovations sequence. Only the single input single output (SISO) case is considered in this paper. After estimating the noise statistics, ISI can be successfully mitigated using the semi-blind SISO equalizer.