A Serial Layered Scheduling Algorithm for Factor Graph Equalization

This letter proposes a novel factor graph equalization based on a serial layered scheduling (SLS) algorithm of belief propagation, where the channel convolutional matrix is decomposed according to the distribution of its non-zero elements. In the SLS algorithm based factor graph equalization, factor nodes are classified into multiple layers referring to the decomposition of channel convolutional matrix, and the a posteriori probabilities message is updated based on a serial layer-by-layer mechanism. This proposed SLS algorithm clearly decreases the signal-to-noise ratio (SNR) threshold and also increases the convergence speed of equalization process, which results in a higher reliability and a lower latency of equalization process. Compared with current parallel flooding scheduling (PFS) algorithm, the convergence speed of SLS algorithm increases almost by twice, and its SNR threshold reduces by 7 dB over the Proakis-C channel.