Direct blind equalization of single-input-single-output (SISO) IIR channel is investigated.Only second-order statistics is required and the channel order need not be known.By oversampling the channel output,we transform the SISO IIR channel into a special Single-Input-Multiple-Output (SIMO) model,in which the subchannels have identical autoregressive (AR) part.A linear prediction approach is applied to the output of this SIMO model and we prove that the prediction error contains the parameters of the first time slot of the model impulse response.These parameters then can be estimated from the covariance matrix of the prediction error.Based on the estimated parameters,zero-forcing equalizers with different delay can be developed.Computer simulation results illustrate the effectiveness of the proposed algorithm.
