Equalization of CDMA downlink channel via Kalman filtering

Abstract

An efficient equalization method for downlink CDMA channels is presented. By representing the transmitted signal as the state of a state-space model, the method employs the Kalman filter (KF) to achieve a minimum-variance unbiased linear estimate of the transmitted chips. The KF is realized at the symbol and chip levels. At the symbol level, the KF is applied to estimation of the transmitted chips that correspond to a particular symbol of interest at each time increment; whereas at the chip level, the transmitted chips are estimated individually. The symbol KF equalizer has a built-in tracking capability that takes advantage of the a-priori known a-periodic scrambling sequence, which renders the transmitted signal non-stationary. In addition, a method for reducing the KF complexity is introduced. Simulations indicate that the KF-based approach is superior to its FIR equalizer counterpart by 1-2 dBs in error-rate performance.