Detection and Mitigation of Known and Unknown DDoS Attacks on Advanced Metering Infrastructure Systems in Nigeria Using Hybrid Machine Learning (Ai) Techniques

Abstract

The enormous rise of network traffic and its diversity on the Internet have posed new and serious obstacles for detecting network attack activity. Distributed denial of service (DDoS) attack is designed to restrict genuine users from accessing a service for an extended period of time. In this attack, the attacker attempts to compromise a large number of hosts to transmit a large volume of traffic to genuine users. Detecting DDoS attacks is difficult and complicated, primarily different DDoS attacks do not common characteristics through which they can be detected. DDoS attacks are very difficult to fight or trace due to their distributed nature, and automated software tools for conducting DDoS attacks are freely available. This paper proposed a DDoS detection model based on hybrid machine learning technique on AMI systems in the Nigeria Utility Business. It has been discovered that detecting unknown DDoS attacks is difficult to analyze as sometimes the IP packets and header are encrypted. This study proposed a combination of Support Vector Machine (SVM) and Artificial Neural Network (ANN) to detect unknown attacks. The AES 256 algorithm has been employed to decrypt the encrypted IP header.

Keywords: ANN, SVM, DDoS attack, AES algorithm, AMI

DOI: 10.7176/NCS/13-04

Publication date:May 31^st 2022