Polar code construction by estimating noise using bald hawk optimized recurrent neural network model.

Polar codes are making significant progress in error-correcting coding due to their ability to reach the limit of the Shannon capacity of communication channels, indicating great advancements in the field. Decoding errors are common in real communication channels with noise. The main objective of this study is to develop a recurrent neural network decoder for robust polar code construction with the Bald Hawk Optimization (RNN-based Decoder with BHO) model that can estimate the error in information bits. This research presents a practical and significant innovation by combining recurrent neural networks (RNNs) for noise estimation in polar coding with a Bald Hawk optimization approach. Moreover, this synthesis of RNN-based noise estimation with Bald Hawk optimization makes the polar coding system more flexible and adaptive, allowing for more accurate noise estimation during decoding. In terms of frame errors, the Bit Error Rate (BER), Binary Phase Shifting Key-BER (BPSK-BER), and Frame Error Rate (FER) achieve the lowest error values of 0.0000087, 0.01519, and 0.000182, respectively. Similarly, in a 4 dB SNR context, the BER, BPSK-BER, and FER achieve values of 0.0000073, 0.02065, and 0.000108, respectively. The results shows that the proposed RNN-based decoder with BHO model outperforms the existing decoders.