Deep Reinforcement Learning Based Bit Flipping Algorithm for Hamming Codes

Hamming codes are effective for single-bit error correction but struggle with multiple-bit errors. While the bit-flipping (BF) algorithm can handle some multiple-bit errors, it also faces limitations. To address these issues, we propose a new decoding method using deep reinforcement learning (DRL) based BF algorithm. Simulation results show that, in a burst channel environment, the proposed method achieves approximately 10% better BER performance compared to the traditional BF algorithm for channel error probabilities ranging from 0.001 to 0.01.