Adaptive LDPC Decoder with Performance Prediction Based on Average Mutual Information

In mobile reception scenario, the channel quality usually shows some fluctuation, while traditional low-density parity-check (LDPC) decoders which utilize fixed decoding algorithms have difficulties to balance the decoding complexity and threshold performance under different channel quality. Adaptive LDPC decoder needs performance prediction in order to guide the selection of decoding options. Traditional performance prediction based on signal-to-noise ratio and channel state information is not precise enough to help the selection of decoding options due to the rapidly varying instant channel state information or incorrect channel-parameter estimation. In this paper, we propose a novel adaptive LDPC decoder with bit-wise average mutual information (AMI) based performance prediction, which could switch between multiple decoding options according to the performance prediction results obtained from the computed AMI and the pre-determined AMI-performance relationship. The proposed decoder adopts multiple candidate decoding algorithms with distinct decoding complexity and threshold performance, and includes the option of direct retransmission. Thus, it could enhance the adaptability of the decoder to the varying channel quality, the robustness to the channel-parameter estimation error, and meanwhile cut down the computing resource waste caused by the avoidable decoding failure. We further propose a novel loop correction method which could effectively correct the channelparameter estimation error by using the decoder output to adjust the decoder input.