A Survey on Intrusion Detection System Using Data Mining Techniques

Nowadays, an increasing number of populations are accessing the Internet for commercial services which is the major cause for attack. Threats are created everyday by an individual or by the organization that attacks the network system. Unusual Malicious activities and unauthorized access are identified by observing the network in Intrusion Detection System. IDS is a passive monitoring system, it warns for suspicious activity but does not prevent them. Various techniques are proposed in IDS to identify attacks or threats in the system. An Intrusion Detection System (IDS) monitors the network for any suspicious or malicious activity and alarms the network administrator for any kind of vulnerabilities. Its main goal is to protect the data’s integrity, confidentiality, or availability of the resources. IDS use data mining techniques to detect intrusions. This survey studies various techniques like k-means clustering, ANN, fuzzy neural network SVM classifiers, AODV based LDK model with AES , k-nearest neighbor classifier, GNS, Naive Bayes, Support vector machines (SVM), J48(C45) with Random Forest, Random Forest with Random Tree Classifiers. All these techniques have been implemented using serial algorithm. IDS have significant overhead in CPU time and memory used by these techniques and so recently researchers are focusing on multi-core CPU to achieve average speedup compared to serial implementation. GPUs act as co-processors which grabs the attention due to their high performance capabilities.