The paper deals with blind identification and equalization of communication channels within the so-called modulation induced cyclostationarity (MIC) framework, where the input symbol stream is modulated by a P periodic precoder with the purpose of inducing cyclostationarity in the transmit sequence. By exploiting the cyclostationarity induced by the periodic precoder, a subspace-based channel identification algorithm that is resilient to the location of channel zeros, channel order overestimation errors, and color of additive stationary noise, is developed. The asymptotic performance of the subspace-based identification approach is analyzed and compared with the asymptotic lower bound provided by the nonlinear cyclic correlation matching approach. Criteria for optimally designing the periodic precoder are also presented. The performance of MMSE-FIR and MMSE-DFE equalizers is quantified for the proposed cyclostationarity-induced framework in terms of the MMSE. Although cyclostationarity-inducing transmitters present several advantages relative to their stationary counterparts from a channel estimation viewpoint, it is shown that from an equalization viewpoint, MIC-based systems exhibit a slightly increased MMSE/BER when compared with the stationary case.
展开▼
