Real Time Intrusion Detection In Edge Computing Using Machine Learning Techniques

The proliferation of edge computing has introduced new opportunities for optimizing latency-sensitive and bandwidth-intensive applications by processing the data closer to its source. In addition, this paradigm shift also brings forth unique security challenges, particularly in the realm of intrusion detection. In edge computing environments, where data is processed at the network edge closer to the data source, real-time intrusion detection is crucial to safeguard the security of the system. The attackers are also exploiting the edge network with rapid extension. Conversely, conventional Intrusion Detection Systems (IDS) cannot detect the latest types of attack patterns in high-speed real-time networks due to their complex behavior and low processing capability. This study introduces a novel approach for developing an effective IDS model to handle such threats in a real-time network and explores the design and implementation of a real-time intrusion detection system (IDS) tailored for edge computing environments. The proposed model is found to be methodical and reliable, and employs supervised Machine Learning (ML) techniques. The objective is to precisely recognize and categorize harmful intrusions or malignant activities within the network in real-time. In order to train and test the model, a self created dataset which utilizes both malevolent and benign PCAPs (Packet Capture files) is used in this research study. To determine the usefulness of the IDS model, the random forest, decision tree, extra tree, and K-nearest neighbors were used as classification techniques. The proposed IDS model exhibhits excellent performance based on several factors such as adaptability and scalability. The model also generates higher values for accuracy, detection rate, F-measure, precision, recall, and lower FPR.