Multi Agent Based Intrusion Detection System Using Neural Networks

The network resources abnormally disrupted by the intruders. Anomaly and signature based intrusion detection schemes are used for the intrusion detection process. Software agent is a reusable component used to perform a preassigned task. Intelligent agent is also one of the software agent based model. The agents perform a task and produce the results to a program or other agent. Intrusion detection steps are handled by a set of agent components. Intelligent agent based intrusion detection system is used for the network environment.

The agent based intrusion detection system uses the signature based approach. All the intrusion signatures are maintained in a database. The network transactions are compared with the signature values. Agent based technique is used in the system. The signature based intrusion detection scheme uses the predefined signatures only. It is not able to detect new attack patterns or behaviors. The signature-based model is enhanced with anomaly-based model. The anomaly based model uses learn the normal user and attacker information from the historical network transaction information.

The intrusion detection system is improved with multi agent model. Transaction analysis, parameter filtering, signature verification, anomaly detection and intrusion detection operations are processed by separate agents. Each agent operations are managed by the system. The neural network technique is used in the intrusion detection process. The back propagation algorithm is used for the anomaly based attack detection process.  The system development is planned with Java language and Oracle back end.

Tags :
Your rating: None Average: 4.5 (2 votes)