A study on applying GA for performance improvement of SP decoder in SMR

We study low-density parity-check (LDPC) coding and iterative decoding methods for shingled magnetic recording in ultra-high-density hard disk drives. Previously, we applied a neural network to evaluate the log-likelihood ratios (LLRs) related to row operations in the sum-product (SP) decoder for LDPC code. Specifically, we updated the LLR considering the influence of noise depending on the recording pattern by providing the LLRs for the decoding target and its adjacent bits to the neural network in SP decoding. In addition, we explored the parameters to update the LLRs by applying the genetic algorithm (GA) to achieve more effective iterative decoding. In this study, to prevent error propagation in SP decoding, we evaluate the LLR for the check node using the parity check information and multiply the LLR for the priori information by the weight to reflect the evaluation result. We also consider the weight applied to the LLRs for the extrinsic information to achieve effective iterative decoding with turbo equalization. Therefore, we employ the GA to efficiently explore the optimal weights to enhance the LLRs for the check node and extrinsic information.