SNR mismatch and robust schemes in MAP-based iterative detection system with CPM

In this paper, two MAP-based demodulation algorithms of CPM modulation scheme are studied. It is shown that these algorithms have almost the same demodulation performance as Maximum Likelihood Sequence Estimation (MLSE) over AWGN channels. However, for the concatenated system followed by a convolutional encoder, the performance of iterative demodulation and decoding using the two demodulation algorithms are widely different. Again, their effects of SNR mismatch on iterative detection scheme are investigated respectively. Simulations show that the optimum MAP detection algorithm is sensitive to SNR estimation errors. Finally, a new iterative detection method is proposed which is not only insensitive to SNR mismatch, but also makes a tradeoff between the BER performance and the robustness to SNR offset. The validity of our way is proved by simulation results.