Minimum MSE relaying in coded networks

In order to combine the advantages of amplify-forward (AF) and decode-forward (DF) in relay networks, several strategies have been developed making use of reliability information after decoding at the relay. The use of soft-output channel decoders enables forwarding reliability information for the decoded bits. This soft information can be forwarded in different ways, e.g. by transmission of the log likelihood ratios (LLRs) normalized to the power constraint. In this paper transmitting expectation values after decoding, the so-called decode-estimate-forward (DEF) scheme, will be shown to be the best choice in terms of the mean squared uncorrelated error at the receiver. Additionally, it will be shown that LLR combining is superior to maximum ratio combining at the receiver as the overall disturbance of the received signal is not Gaussian in the case of DEF. Furthermore, in the uncoded case LLR combing also improves the performance in terms of effective signal-to-noise ratio and bit error rate.