A Novel Framework for Intrusion Detection Using Distributed Collaboration Detection Scheme in Packet Header Data

In recent years, the number of attacks on the computer network is voluminous. Secure data communication over the network is always under threat of intrusions. To protect from these attacks various intrusion detection techniques have been developed. Anomaly detection system detects the novel attacks based on deviation of the behavior of packets from the normal flow and Signature detection system detects known attacks based on stored signatures. We have proposed a Distributed collaboration detection scheme that combines the advantages of Anomaly and Signature based method by capturing the packets in real time. The uninteresting traffics are filtered by packet filtering and further normalization. The relevant features are selected based on our Correlation based BAT Feature Selection (CBBFS) Algorithm. Our Proposed Efficient Behavioral Prediction (EBP) scheme analyzes the episodes and classifies the attack based on EGSSI. Then Proficient Ordinance Generation (POG) for Inspection of IP Phase labels the IP as trusted or untrusted. Our proposed framework outperforms the results of existing classification algorithms (C4.5, Naive Bayes, PSO, GSA and EDADT) by reducing the rate of false positives.