Performance analysis of blind channel estimation for precoded multirate multiuser systems

We have proposed a blind channel estimation method for a precoded multirate multiuser communication system. The method requires subspace decomposition of a data covariance matrix. In this paper we analyze the method in the presence of imperfect estimation of the covariance matrix due to finite data sample effect. Based on perturbation theory, the first order perturbation of the noise subspace is derived. Then the covariance and the mean square error of the channel estimates are obtained in closed forms. They explicitly depend on the statistical properties of the input and the additive noise up to the fourth order. Therefore, they can be evaluated for different scenarios.