A Hybrid Approach for Intrusion Detection System using Data Minining and Artificial Neural Network

This paper is going to present a hybrid model for IDS using ANN and data mining approaches. it is important to deploy a hybrid network using ANN based and data mining based IDS systems that can enable the system in making decisions on intrusion in a mobile environment. MANET is an infrastructure-less ad-hoc network. It is a collection of mobile nodes that are connected in an arbitrary and dynamic manner. ANN (Artificial Neural Network) is a type of Machine Learning scientific and statistical model that is used by computer systems in order to perform a task without involvement of explicit intrusion, rather than relying on interference and patterns instead. As there is node mobility that can impose different security issues, MANETs are found more susceptible in security provision. In order to resolve this security issue, some authentication and encryption techniques can be proposed for the first-line defensein order to mitigate the security risks. However, complete eradication of these types of risks is next to impossible. For theSecond line defense, the necessity of an Intrusion Detection System (IDS) is essential in this case. IDS can be referred as a method tool or resource that can help to detect access and warn for any activity of unapproved or unauthorized network activity. The current research has been done to prevent the network from the intruders. The number of attackers has been increased in the network because of more usage of technology and wireless network. Data is not allowed by the attackers to transmit successfully from source node to destination node. These intruders want to steal data and let to the unsuccessful transmission. In this research we are deploying 48 to 66 nodes in a defined communication area of 1000*1000 height and width. While starting the transmission source node and destination node is defined. After that while using IDS and AODV protocol route is created between destination and source node. If network decreases its performance will need to apply cuckoo search as an optimization algorithm. Cuckoo search algorithm extracts the properties and observe the energy been consumed by each node. Optimized properties provided as an input to the decision tree classifier with the help of this, system is being prepared according to their properties. If there will be a chance for existence of intruder in the network then it will be classified using decision tree algorithm. At the end QOS parameters energy consumption is performed using cuckoo search optimization algorithm. The detected intruder is prevented using decision tree and removed from the MANET network.