A two-stage blind identification algorithm for CDMA time-varying dispersive channels
Peng Qi-cong · Journal of China Institute of Communications · 2002
In this paper, we developed the discrete-time canonical model of CDMA time-varying channels by sampling the channel抯 spread function in time-frequency domain. A new computationally efficient two-stage blind identification algorithm was proposed for this equivalent model. In order to cancel error propagation effects of this new method, we converted the ideal single-input multi-output (SIMO) model into an additive Gaussian noise model. Unlike the basis expansion approach, the new blind algorithm doesn抰 require estimation of the basis frequencies based on the higher-order statistics (HOS) of the observations. Simulation demonstrates the effectiveness of the algorithm.