Intrusion Detection System using Signature and Anomaly based Algorithm

With the rise of sophisticated cyber threats, securing networks and protecting sensitive information requires advanced security mechanisms. Intrusion Detection Systems (IDS) play a crucial role by monitoring network traffic, detecting unauthorized access, and identifying malicious activities. Traditional *signature-based IDS are effective against known threats but often fail to detect zero-day attacks. In contrast, anomaly-based IDS utilize machine learning to identify deviations from normal behavior, enabling the detection of emerging and unknown threats. This project proposes a hybrid deep learning-based IDS that combines Convolutional Neural Networks (CNN) for hierarchical feature extraction with Fully Connected Neural Networks (FCNN) to enhance detection accuracy and performance. The model is trained and evaluated using the UNSW-NB15 dataset, with an 80:20 train-test split after preprocessing to ensure reliable performance. Deployment is done using Flask, enabling real-time intrusion detection through an interactive user interface. The proposed system addresses common IDS limitations such as high false positive rates, evasion techniques, and scalability challenges. By combining signature-based detection for known threats and anomaly-based detection for novel behaviors, this hybrid model improves overall detection capabilities. Future enhancements may involve integrating automated threat intelligence, expanding hybrid detection strategies, and incorporating real-time behavioral analytics. Leveraging AI and deep learning, this project contributes to developing more proactive and adaptive IDS solutions capable of countering evolving cyber threats.