Data-aided SIMO channel estimation with unknown noise spatial covariance matrix

摘要

In this letter, we investigate the channel estimation for single-input multiple-output systems, where the channel vector and the noise spatial covariance matrix (SCM) are jointly estimated. By utilizing the inherent relationship of the received data symbols’ sample covariance matrix, the SCM and the channel vector, we propose a conditional maximum likelihood (CML) estimator, which is the solution of a non-convex optimization problem. The global optimum can be expressed in a quasi-closed-form with one unknown scalar parameter, which can be efficiently identified via solving polynomial equations. Simulations show that the proposed CML estimator achieves significant performance gains compared with the traditional maximum likelihood estimator, as long as the data symbol number is not too small.