Improvement of Network Traffic Prediction in Beyond 5G Network using Sparse Decomposition and BiLSTM Neural Network

Companies providing telecommunication services, especially in Beyond 5G networks, are increasingly interested in traffic forecasting to improve the services provided to their users. However, forecasting network traffic is challenging due to traffic data's dynamic and non-stationary nature. This study proposes an effective deep learning-based traffic prediction technique using BiLSTM (Bidirectional Long Short-Term Memory). The proposed method begins with preprocessing using K-SVD (K-means Singular Value Decomposition) to reduce dimensionality and enhance data representation. Next, sparse feature extraction is performed using Discrete Wavelet Transform (DWT), and a sparse matrix is constructed. A Genetic Algorithm (GA) is used to optimize the sparse matrix, which effectively selects the most significant features for prediction. The optimized sparse matrix is fed into the BiLSTM model for accurate traffic forecasting. Experimental results show that the proposed method significantly reduces Mean Absolute Error (MAE), Mean Squared Error (MSE), and Root Mean Squared Error (RMSE) while achieving higher accuracy (ACC) compared to traditional neural networks. The results demonstrate that the proposed sparse matrix, integrated with BiLSTM, provides superior prediction accuracy and better generalization, making it a robust solution for network traffic forecasting in Beyond 5G networks.