Jose, Angel Maroor (2024) Artificial intelligence-based Cache Partitioning for protecting the systems against vulnerabilities. Masters thesis, Dublin, National College of Ireland.
Preview |
PDF (Master of Science)
Download (1MB) | Preview |
Preview |
PDF (Configuration Manual)
Download (532kB) | Preview |
Abstract
Efficient cache management is vital in contemporary computer systems to optimize performance by minimizing latency and maximizing efficiency. Cache partitioning is a method that distributes cache resources across many processes or threads to guarantee equitable and effective consumption while reducing cache congestion. The goal of this project is to create a new cache partitioning method that utilizes artificial intelligence (AI) to improve cache use. This will result in higher hit rates and lower access latency. The suggested approach observes cache behavior and implements partitioning exclusively when abnormal or significant cache patterns are identified, therefore reducing RAM usage and releasing capacity. The system selectively intervenes by utilizing machine learning models to identify detrimental cache behaviors, resulting in improved speed and security. Out of the different models assessed, XGboost, Random Forest and DNN had the greatest performance, obtaining a 93% accuracy across all evaluation criteria. The paper further examines existing research on cache partitioning, vulnerability identification, and the use of artificial intelligence in the field of cybersecurity. The results emphasize the capability of this method to enhance the efficiency and safety of a system in intricate, multi-threaded settings.
Actions (login required)
![]() |
View Item |