Channel shortened chip-level MMSE equalization for CDMA downlink

Previous work has shown that MMSE (Minimum Mean Square Error) equalizer has better performance than RAKE receiver and ZF (Zero Forcing) equalizer in high data rate systems. However, MMSE equalizer has a much higher complexity than RAKE receiver, especially when fading channel is sparse in high chip rate cases and the length of the channel becomes very long, which prohibits its applications. Our purpose of the paper is to reduce the length of the channel efficiently and thus to reduce the computation complexity of MMSE equalizer. A new algorithm for the optimal shortening of channel impulse response for MMSE equalizers is proposed. This algorithm, which uses eigenvalues and eigenvectors to generate the globally optimal coefficients of Shortening Impulse Response Filter (SIRF), has a lower complexity than the previous ones.