Anomaly-based network intrusion detection through assessing feature association impact scale

摘要

Phenomenal growth in the volume of computer network users leads to the drastic divergence of anomaly activities. Henceforth, it is quite obvious to consider the associability between network transactions and the feature involved to form those transactions. In this regard, the majority of current research is involved to devise signature-based intrusion detection using softcomputing techniques. Most of these soft computing approaches are delivering the computational complexity as O(n~2), which is due to magnification of number evolutions. Here in this paper, a meta-heuristic statistical scaling process is derived to estimate if a network transaction is safe, suspicious or intrusion. The proposed model is using duplex graph strategy to estimate the strong associability of the features towards network transactions. The results explored from the empirical study are successfully delivering the accuracy towards identifying the intrusive state of a network transaction. The proposed strategy is able to stabilise the computational complexity to O(n∗log(n)) towards assessing the network intrusion using the scale derived.