Attack Detection in a Rule-Based System Using Fuzzy Spiking Neural P System

Abstract

The virtual area of communication known as cyber space brought about by the debut of the internet has enabled some cyber crimes – ‘Intrusion’ inclusive. So, efforts are being geared towards ensuring that reliable and efficient Intrusion Detection Systems (IDSs) are developed to curtail this menace. However, Spiking Neural P (SN P) systems have been established as a class of distributed parallel computing models. So, in this paper, a novel network intrusion prediction model based on trapezoidal Fuzzy Reasoning Spiking Neural P (tFRSN P) system, is implemented for the very first time for the detection of intrusion. tFRSN P system is an extension of SN P system. It has a graphical modeling advantage which makes it well suited for fuzzy reasoning as well as fuzzy knowledge representation using If-Then rules. The dynamic firing power of neurons is harnessed in a simple parallel matrix-based fuzzy reasoning format to generate inferences. To establish the effectiveness of this approach especially in the area of speed of parallel reasoning and the handling of uncertainties, detection of Brute Force Attack (BFA) is used for demonstration. From the crisp results (0.0431, 0.0414, 0.4453, 0.1703 and 0.0414) obtained, it shows that the parallel processing capability of tFRSN P system could be used to rapidly reason and analyze the severity (possibility of an attack) from any network data.