Novel SNR Estimation for Turbo Decoding over Rayleigh Fading Channels

It is known that the signal to noise ratio (SNR) mismatch may cause degradation in error correction performance of turbo decoding. In order to obtain desired coding gain, a turbo decoder can perform online SNR estimation from the received data in each code block. This paper presents a novel SNR estimation method for turbo decoding over Rayleigh fading channels by utilizing the estimate ratio of the fourth moment to the fourth power of mean absolute value over the received data. Analysis demonstrates that the proposed method is superior to the previous method in the estimation accuracy, especially when code block length is small. Furthermore, simulation shows that nearly optimal bit error rate (BER) can be achieved even when the code block length is reduced to 400 bits (200 information bits, half code rate), while the previous method results in noticeable BER degradation