Transudative Long Short Term Memory based Network Traffic Prediction in Wireless Mesh Networks

In Wireless Mesh Network (WMN), traffic-conscious consistent entrance selection is significant for improving the Quality of Service (QoS) and enhancing the effectiveness of the network model. However, the complex spatial-temporal patterns of the user demand concentrate on the prediction of the mobile traffic, which accounts for a majority of the present network, thereby being challenging. Hence, this research proposes the Transudative Long Short-Term Memory (TLSTM) approach for the prediction of the Network traffics in WMN. The TLSTM model's ability to capture long-term dependencies helps in understanding and predicting complex traffic patterns and trends. Initially, the data is collected from High-Speed Diesel (HSD) pump, which is predicted through detecting the model’s results. The effectiveness of the HSD pumps is detected through a vibration parameter as an outcome parameter. The experimental results show that the proposed TLSTM approach attains a better Mean Absolute Error (MAE) of 0.008 and 0.009 respectively as the monthly and yearly prediction as compared to the existing method, hybrid Convolutional Neural Network with LSTM (Conv-LSTM) approach.