Network Traffic Anomaly Detection Algorithms on Distributed Systems using Cognitive Intelligence

Network traffic monitoring is one of the important roles to maintain the security and confidentiality between distributed systems particularly in detecting early cyber threats. Distributed system is a large network interconnected device, which are connected with one another. So, detecting anomalies is a major challenge in these systems. Traditional systems fail to detect anomalies in early stages, because threats are too advanced which are not handled by the traditional capacities. To address this issue the present study proposed and improved version network traffic monitoring system called (Net-IV). This approach combines the advantages of 1D-CNN, Long Term-Short Term Memory (LSTM) and GRU (Gated Recurrent Units). According to this, 1D-CNN which is well known for its feature extraction ability, whereas LSTM helps to analyse the temporal dependencies, finally GRU refine the overall performance and helps to detect anomalies with greater precision. The model was evaluated using CIC-1DS-2017 dataset, a complete benchmark dataset for intrusion detection system. Through the simulation, we observed that the suggested Net-IV achieves a remarkable accuracy rate of 99.78% and F1-Score of 99.56% which is 0.05% higher than the existing DCGCANet model. Thus, the results suggested that the proposed Net-IV system could be effectively installed in real-time, to protect the distributed system confidentialities from various forms of cyber-attacks.