Recursive blind identification of non-Gaussian time-varying ARmodel and application to blind equalisation of time-varying channel

摘要

A novel method for the blind identification of a non-Gaussiantime-varying autoregressive model is presented. By approximating thenon-Gaussian probability density function of the model driving noisesequence with a Gaussian-mixture density, a pseudo maximum-likelihoodestimation algorithm is proposed for model parameter estimation. Thereal model identification is then converted to a recursive least squaresestimation of the model time-varying parameters and an inference of theGaussian-mixture parameters, so that the entire identification algorithmcan be recursively performed. As an important application, the proposedalgorithm is applied to the problem of blind equalisation of atime-varying AR communication channel online. Simulation results showthat the new blind equalisation algorithm can achieve accurate channelestimation and input symbol recovery