EECS MSc Students, Sasidharan Velu and Subramaniam Subramanian Murugesan, have published their new research in Springer Cluster Computing (Q1 Journal with IF: 4.4).
Researchers from the School of Electronic Engineering and Computer Science (EECS) at Queen Mary University of London have recently developed a new testbed for Artificial Intelligence (AI)-Based Cloud Computing environments under the supervision of Dr. Sukhpal Singh Gill and published it as a research paper titled “CloudAIBus: a testbed for AI-based cloud computing environments”. The study, available in the “Springer Cluster Computing” journal, presents a novel AI-driven approach to optimize resource allocation in cloud computing environments. This research addresses the significant challenge of resource overprovisioning in cloud computing. Traditional resource allocation strategies often lead to financial inefficiencies due to overprovisioned CPU resources. The team created CloudAIBus, a testbed that uses deep learning techniques to provide accurate resource usage forecasts, allowing for cost-effective and efficient resource allocation.
Key Features and Innovations
CloudAIBus employs a deep learning model named DeepAR, implemented using Amazon SageMaker, to predict CPU usage and make informed resource allocation decisions. The model was trained and tested using the Google Cloud Platform, demonstrating its superior performance over traditional approaches such as Long Short Term Memory Loss (LSTM) and Autoregressive Integrated Moving Average (ARIMA). The testbed significantly reduced unused CPUs from 98.65% to 32.35%, highlighting its effectiveness in minimizing resource wastage.
Research Impact
This innovative approach not only enhances cloud computing efficiency, but it also aligns with global sustainability efforts by reducing the environmental footprint of cloud services. The study's implications extend to various sectors, including finance, retail, and supply chain management, where precise resource forecasting is crucial. The publication of this research marks a significant milestone in the field of cloud computing, showcasing the potential of AI to revolutionize resource management.
Software Availability
The researchers have released code as open-source software. The implementation code with experiment scripts and results can be found at the GitHub repository. For further information, watch this video on YouTube.
Publication Details
Sasidharan Velu, Sukhpal Singh Gill, Subramaniam Subramanian Murugesan, Huaming Wu, Xingwang Li. CloudAIBus: A Testbed for AI based Cloud Computing Environments. Springer Cluster Computing, 2024. DOI: https://doi.org/10.1007/s10586-024-04562-9
Read the full publication