Research
My research focuses on distributed systems, with special interest in virtual machine fault tolerance, anomaly detection in cloud computing environments, and cloud resource allocation optimization algorithms.
Current Research Projects
Optimal Destination Node Selection in Live Virtual Machine Migration
June 2024 - Present
This research addresses the NP-Hard problem of finding optimal destination hosts for Virtual Machines in Cloud Datacenters. By utilizing an Ant Colony Optimization algorithm, we aim to develop a solution that efficiently places VMs while considering multiple constraints including resource utilization, energy efficiency, and SLA requirements.
Key Research Questions
- VM placement optimization for efficient resource utilization
- Effective heuristics for ant colony optimization algorithm
Methodologies
- Ant Colony Optimization
- CloudSim Plus simulations
- Statistical performance analysis
Virtual Machine Proactive Fault Tolerance using Log-based Anomaly Detection
March 2023 - April 2024
This research focused on identifying potential Virtual Machine failures due to hardware/software faults in real-time using an anomaly detection approach based on our improved Matrix Profile algorithm. The system analyzes system logs to detect patterns indicative of impending failures, allowing for proactive migration before service disruption occurs.
Publications
Virtual Machine Proactive Fault Tolerance using Log-based Anomaly Detection
Pratheek Senevirathne, Samindu Cooray, Jerome Dinal Herath, and Dinuni K. Fernando
IEEE Access (2024)
This paper presents VMFT-LAD, a semi-supervised real-time log anomaly detection model that combines our modified Matrix Profile algorithm with Large Language Models to enable proactive virtual machine fault tolerance in cloud computing environments. We demonstrate that our approach achieves exceptional performance with a 96.28% early detection rate and minimal false positives, allowing for timely VM migration before failures occur without requiring labeled failure data.
DOI: 10.1109/ACCESS.2024.3506833