Iterative Error Decimation for Syndrome-Based Neural Network Decoders

In this letter, we introduce a new syndrome-based decoder where deep neural network (DNN) estimates the error pattern from reliability and syndrome of received vector. The proposed algorithm works by iteratively selecting most confident positions to be bits pattern, updating vector when position is selected. Simulation results for (63,45) (63,36) BCH codes show that approach outperforms existing decoders. addition, flexible in it can applied on top any DNN without retraining.