Turbo-decoding without SNR estimation

Theoretically, it is necessary to estimate the SNR when using a MAP of the log-MAP constituent decoder. The effect of an SNR mismatch on the bit error rate performance of turbo-codes and the design of good variance estimators have been addressed by several authors. In this letter we study the SNR sensitivity of turbo-decoding with log-MAP and max-log-MAP constituent decoders respectively for AWGN and Rayleigh fading channels. Our theoretical and simulation results indicate that an estimation of the SNR is not necessary from a practical point of view. Our setup is aligned with decoder implementation aspects of future mobile communication systems.