An Improved Approach Based on Density-Based Spatial Clustering of Applications with a Noise Algorithm for Intrusion Detection

Abstract

Network Intrusion detection systems (NIDS) are extremely important for make the network secure from unauthorized access. Numerous studies have already been conducted to detect the unauthorized access to achieve security. As the NIDS are still lacking in terms of accuracy, true positive rate (TPR) and the false positive rate (FPR) of the invasive events. The main cause of high FPR in intrusion detection systems is run with a default set of signatures. Issues in the detection rate are caused by feature similarities between man-made events and environmental events. Considering this fact, in this paper, we introduced a new intrusion detection algorithm named as I-DBSCAN by focusing on the above-mentioned issues to get the better results from the previously done experiments. We used clustering and classification techniques. The proposed algorithm is an enhanced version of the existing DBSCAN algorithm. However, this research can spot attacks on data from IDS. It is found that the novel algorithm achieved more accuracy when it is applied to four classification methods on KDD Cup 99 and NSL-KDD Cup99 data. The results of our proposed methodology are more efficient with the achievement of better accuracy level and false positive rate (FPR).