International Research Journal of Engineering and Technology (IRJET)
e-ISSN: 2395-0056
Volume: 11 Issue: 08 | Aug 2024
p-ISSN: 2395-0072
www.irjet.net
DESIGN AND IMPLEMENTATION OF AN ENHANCED MODEL OF SUPPORT VECTOR MACHINE WITH RADIAL BASIS FUNCTION AND PARTICLE SWARM OPTIMIZATION FOR DETECTION OF DDoS ATTACKS Anthony Obogo Otiko1, Emmanuel A. Edim2, Emmanuel U. Oyo-Ita3 1’3Department of Computer Science, University of Cross River State, Calabar. 2Department of Computer Science, University of Calabar
-----------------------------------------------------------------------***--------------------------------------------------------------------With the expansion of services facilitated by ICT, including Abstract:
the management of Critical Infrastructure (CI), society's dependency on these systems has grown substantially. Any disruption to these systems, even for short durations, can have far-reaching consequences across various facets of modern life.
The increasing reliance on networked systems in the Information and Communication Technology (ICT) landscape has made them crucial for managing Critical Infrastructure (CI) and information access. However, this dependence has also made these systems vulnerable to Distributed Denial-of-Service (DDOS) attacks, which disrupt services by flooding target hosts with traffic. Traditional security measures often fall short against these attacks due to their dynamic nature and evolving tactics. To address this challenge, various researchers have explored the potential of Machine Learning (ML). This study proposes a hybrid solution that combines a Support Vector Machine (SVM) with a Radial Basis Function (RBF) kernel and Particle Swarm Optimization (PSO) for feature selection to enhance DDOS attack detection accuracy. Using the KDD-99 dataset, the SVM-PSO model is trained and evaluated, achieving a superior performance of 99.99% compared to SVM-RBF alone with 99.72%. The findings demonstrate the efficacy of the proposed approach in detecting DDOS attacks, showcasing its potential for bolstering cyber defences. Additionally, comparative analysis with state-of-the-art algorithms highlights the superior accuracy of the SVM-RBF-PSO model, underscoring its effectiveness in mitigating DDOS threats. This study contributes valuable insights into enhancing network security through innovative machine learning techniques, addressing the persistent challenges posed by DDOS attacks.
Distributed Denial-of-Service (DDOS) attacks, in particular, have emerged as a significant threat in the digital realm (Kaur et al., 2019). These attacks leverage distributed computing power to inundate target hosts with traffic, leading to service disruptions. The proliferation of DDOS attacks, coupled with the limitations of traditional security measures, has made them a persistent challenge for network defenders. Despite efforts to mitigate DDOS attacks, their dynamic nature and evolving tactics pose ongoing challenges for detection and prevention. Traditional security measures often prove insufficient against the scale and sophistication of modern DDOS attacks, necessitating innovative approaches to bolster cyber defenses (Shukla et al., 2022). Hence, diverse researchers have applied the viabilities of Machine Learning (ML), knowledge-based approaches, and statistical analysis to tackle the problem of DDOS attacks. However, each method presents its unique set of limitations and challenges. For instance, statistical methods often struggle to precisely determine the normal distribution of network packets (Sharafaldin et al., 2019). While ML techniques offer promising results, however, the optimal selection of best features remains a primary concern considering the prodigious size of datasets.
Keywords: Algorithm, artificial intelligence, Denial of
Service, Distributed Denial of Service, Kernel PCA, Support Vector Machine, Malicious Traffic.
While traditional Feature Selection (FS) methods have demonstrated viability, their efficacy remains subject to scrutiny (Khaire and Dhanalakshmi, 2022). Notably, recent trends underscore the increasing adoption of Metaheuristic approaches, particularly as wrapper methods for feature selection, showcasing substantial effectiveness in mitigating the impact of overwhelming dataset sizes within
1.0 Introduction In the rapidly evolving landscape of Information and Communication Technology (ICT), the reliance on networked systems for information access and communication has become paramount (Agboola, 2023).
© 2024, IRJET
|
Impact Factor value: 8.226
|
ISO 9001:2008 Certified Journal
|
Page 412