Statistical Implementation for SD-RNN Model for Multi-Class Classification for Network Intrusion Detection System

Abstract

With the increasing popularity of technologies that depend on computer networks, providing security is of paramount importance. Consequently, intrusion detection systems (IDS) play a vital role in monitoring these networks. An essential element in securing the network of an organization is the Intrusion Detection System (IDS), which serves as the primary barrier against cyber threats and is tasked with thwarting illegal entry into the network. Flow-based network traffic analysis is commonly used in IDS solutions to identify security concerns. In recent years, several innovative strategies have been proposed and implemented to address the issue of network security, with a specific focus on Intrusion Detection Systems (IDSs). The creation of intrusion detection systems using AI methods is a modern method for finding breaches in a network. Since there are many possible approaches, it is important to have a standardized method that facilitates good judgement when classifying intrusions. This paper presents a novel approach utilizing deep learning and machine learning modelling to develop a model for multi-class classification. The principal objective is to create IDS that can effectively detect anomalies using flow-based analysis. The latest CICIDS2017 dataset is used in experimentation for testing and training. The experiments performed with a deep learning model for IDS produced promising results, achieving a 99.77% accuracy rate for multi-class classification while employing the specific dataset.