Network Performance Monitoring and Bottleneck Detection Using Improved Recurrent Neural Network

Monitoring network performance and bottleneck detection assist in identifying slowdowns, congestion points, and ineffectiveness in network traffic flow. Bottleneck detection helps to identify misconfiguration, overloaded links, and faulty performance which allow timely corrective actions. Therefore, ensuring smooth network performance is significant for managing security, reliability, and user experience. However, identifying bottlenecks in network and system performance is difficult because of the complexity of traffic patterns leads to inaccurate performance analysis. Therefore, this research proposes improved Recurrent Neural Network (RNN) for monitoring network performance and detecting bottleneck accurately by resolving sequence learning issues. The proposed improved RNN is evaluated using N_BaIoT dataset to determine the model performance. Then, the obtained data is pre-processed using min-max normalization to standardize data and one-hot encoding to encode the labels effectively. At last, the proposed improved RNN achieves a better accuracy of 99.99% accuracy compared to existing methods like hybrid Deep Neural Network (DNN) and Long Short-term Memory (LSTM) using N_BaIoT dataset.