Low-complexity channel estimation for beyond 3G systems

A robust Minimum Mean Squared Error (MMSE) channel estimation algorithm that does not rely on a priori knowledge of the channel statistics gives good results with MC-CDMA systems. The robust MMSE estimator takes advantages of the correlation between all Np pilot-subcarriers and requires one Np x Np matrix inversion per each pilot symbol of the burst structure. This complexity can be large depending on the number of pilots subcarriers in the system. This paper presents and analyses two low-complexity suboptimal approximations of the robust MMSE channel estimator. A complexity versus performance degradation comparison is done. The performance is presented in terms of Mean Square Error for a 1024 tone MC-CDMA system over ETSI BRAN mobile channel models.