Rake Reception With Channel Estimation Error

Several digital cellular systems employ coherent RAKE reception, wherein channel estimates are used to combine despread values. In this paper, a maximum likelihood (ML) approach to RAKE combining that accounts for channel estimation error, including error correlation across despread values, is examined. Fading and noise correlation are also considered. The performance of the ML approach is compared to traditional RAKE combining as well as approximate ML-based approaches. Results show that when the channel estimation error and noise are of the same order, ML-based approaches can provide gains on the order of 1 dB over traditional RAKE reception