A Bayesian Approach to Adaptive RAKE Receiver

Abstract

The RAKE receivers are widely used in code division multiple access communication systems to achieve anti-multipath fading. However, a traditional RAKE receiver requires pilot data stream inserted into the sequence, which occupies channel resources and limits its applications. In this paper, a new adaptive RAKE receiver based on Bayesian theory is reported that only uses received signals to estimate its channel parameters. Observed data are used to obtain the information of the channel impulse response. Next, the prior information is used. In the iterative process, a priori information is accumulated to improve the receiver performance. Thus, the mean and covariance of the channel impulse response that is modeled as a complex and uncertain Gaussian random vector are recursively estimated using Bayesian theory. Finally, the RAKE weights are obtained using the mean and covariance. As shown in the simulation results, the bit error rate (BER) decreases as the number of fingers increases. The performance of the new RAKE receiver has been greatly improved compared with the all-RAKE receiver with maximal ratio combining, RAKE receiver with singular value decomposition, and RAKE receiver with fast approximated power iteration. Under medium to high SNR conditions (i.e., ≥−5 dB), the BER performance of the new RAKE receiver provides at least $3\times 10^{-4}$ less than that of the other receiver tested.