A method of estimating channel impulse response in a signal transmitted over a channel in a communication system is described. The channel estimator obtains a priori knowledge about the transmitted signal and then use the transmitted signal and the a priori knowledge to choose an estimate of channel impulse response which minimises the expected distance between the transmitted signal and a reconstructed signal. The expected distance that is minimised is a cost function, represented by Eä PARALLEL r - Bh PARALLEL <2>¦rü. By using soft decision feedback, the invention minimises erroneous decision feedback which can cause error propagation. The decision are usually in the form of log likelihood ratios (LLR). <IMAGE>