DESIGN AND APPLICATION OF MACHINE LEARNING METHODS FOR CLOUD SECURITY AND DDOS ATTACK DETECTION

DOI: https://doie.org/10.0126/Jbse.2025695699

Amarnath J L, Dr.Chandramouli H , Dr. Pritam G Shah , Dr. I Manimozhi

Keywords:

Distributed Denial of Service, DDoS, Cloud Computing, Security and Machine Learning.

Abstract:

Cloud computing is a paradigm that delivers software and hardware services over the internet. By adhering to the principles of cloud computing, users can access and manage data and applications from any device connected to the web. In today's digital age, it is almost unimaginable for an individual to be without internet access, even if they could live without a specific gadget. The advantages of cloud computing include scalability, virtualization, user accessibility, lower infrastructure costs, and flexibility. However, one significant drawback is its vulnerability to distributed denial of service (DDoS) attacks. These attacks involve multiple computers simultaneously targeting a specific resource, website, or server to deny service to end users. The onslaught of fraudulent connection requests and an abnormal volume of messages and malformed packets can slow down or shut down the system entirely, preventing legitimate users from accessing the services they need. This article explores the application of machine learning algorithms to detect DDoS attacks. Two primary techniques were employed in this research using datasets from the NSL KDD repository. On one hand, we utilized the Learning Vector Quantization (LVQ) filter; on the other hand, we applied Principal Component Analysis (PCA), a dimensionality reduction technique. For detecting DDoS attacks, we categorized the characteristics from each approach using Decision Trees (DT), Naïve Bayes (NB), and Support Vector Machines (SVM). We then compared the outcomes of these different categorizations. The results showed that LVQ-based DT outperformed other types of DT in identifying attacks effectively.

PDF