Robust estimation of the linear model parameters in the presence of moving average noise

The parameters of a linear model in the presence of dependent perturbations such as the moving average can be estimated using a range of whitened $M$-estimates which are extenstions of the Huber $M$-estimates. The consistency and normalized consistency of the proposed estimates are proved with general assumptions on perturbation model input vectors distribution. Perturbations whose variance is infinite are admissible; consequently, data scatter can be recognized. The model input vectors can be nonuniform and dependent.