Mobile Ad-hoc Networks (MANETs) have no clear line of defense; and therefore, beside legitimate network nodes, they are also accessible by malicious nodes. Traditional ways of protecting the network (such as firewalls) are not sufficient and effective. Therefore, intrusion detection systems (IDS) are required to monitor the network and detect the misbehavior and anomalies. Intrusion detection is the act of detecting actions that attempt to compromise the security goals. Intrusion detection systems encounter challenges such as misdetection, misjudgment, and slow response to the attack. In recent years, several data mining techniques as classification, clustering, and association rule discovery are being used for this purpose. In this paper, we propose a hybrid technique that combines data mining approaches like K-Means clustering algorithm and AFS theory as a feature selection module. The main purpose of the proposed technique is to decrease the number of attributes associated with each data point. The proposed technique performs better in terms of detection rate and accuracy when applied to KDD CUP 99 dataset in comparison with other intrusion detection systems in the detection of DoS and Probe attacks.
