Design and Development of Novel Error Detection and Correction in Communication Systems

This research presents the development and performance evaluation of a Hybrid Convolutional Neural Network-Assisted Decoder (HCNAD) for error detection and correction in digital communication systems. The proposed method combines the classical Golay coding scheme with neural network-based decoders, specifically Multi-Layer Perceptron (MLP) and Convolutional Neural Network (CNN) architectures. Comparative analysis was performed with the traditional Golay baseline using performance metrics such as Bit Error Rate (BER) and Frame Error Rate (FER) under varying signal-to-noise ratios (Eb/N₀). Experimental results demonstrate that the CNN-assisted decoder consistently outperforms both the baseline and MLP-assisted models, especially in low signal-to-noise conditions. The findings highlight the potential of integrating deep learning with conventional coding schemes to enhance communication reliability and reduce decoding errors in noisy environments.