Heuristic Approach of Supervised Learning for Intrusion Detection

The objective of the study is to design a heuristic supervised learning algorithm, that locates Intrusion and improves performance of network. In this paper we have proposed the algorithm using existing supervised learning approach and evaluated IDS for MANETs. The trained dataset of KDDCUP is loaded and known four attacks are taken for evaluation. We assume that discussed four attacks dominate the network traffic. The algorithm is iterative in nature to produce optimum results. The performance of proposed supervised algorithm is evaluated under different network traffic and mobility patterns for Dos, PRB, R2L and U2R attacks. The results indicate high accuracy for almost all the four attacks. The proposed algorithms, results show high accuracy on discussed four attacks of KDD 99. It also produces low high positive rate.