Context-aware Workload Optimisation in Cloud Ecosystems: A Framework for Sustainable and Low-Impact AI Processing Kingston University in United Kingdom

This PhD project at Kingston University addresses the urgent need for sustainable and energy-efficient cloud computing, particularly in the context of artificial intelligence (AI) and big data processing. As cloud-based AI applications continue to grow, so does their environmental impact due to high energy consumption. The research aims to develop a context-aware framework for workload optimisation, focusing on dynamic workflow management and resource allocation to balance computational demand and energy usage.

Unlike traditional methods that prioritise computational speed or rely solely on renewable energy, this project integrates contextual factors such as task priority, workload characteristics, and node availability to optimise task scheduling and resource allocation. By leveraging federated computing and modular architectural approaches from software engineering, the research will enable intelligent distribution of workloads across nodes or server farms, preventing overload, reducing idle energy consumption, and enhancing overall resource efficiency. These strategies are particularly suited to decentralised, scalable cloud systems, offering new opportunities for building sustainable digital infrastructures.

Modular architectures are central to the project, providing flexibility and adaptability in workload distribution. Systems can be divided into smaller, independently manageable components that are activated or scaled according to specific task requirements, minimising unnecessary energy use. The integration of modular principles with federated computing—distributing computational tasks across geographically dispersed nodes—will be explored to optimise workload distribution, route tasks to underutilised resources, and lower energy costs.

The research also incorporates event-driven architectures, which allow systems to respond to specific triggers and eliminate continuous polling, further reducing idle resource consumption. Serverless computing paradigms, where cloud providers dynamically manage resources, will be examined for their potential to allocate resources only as needed, ensuring efficient execution of AI and big data workloads while minimising energy waste.

Methodologically, the project combines theoretical and practical approaches, aiming to produce context-sensitive algorithms for resource allocation and workload scheduling that balance energy efficiency with computational performance. The expected outcomes include innovative solutions for reducing the carbon footprint of cloud-based AI and big data processing, contributing to the broader field of sustainable computing.

Ideal candidates will have strong programming skills (Python, Java, or C++), a solid understanding of software engineering principles (including modular and event-driven systems), and familiarity with cloud computing platforms and federated computing frameworks. Analytical skills for evaluating energy efficiency and performance metrics using simulation tools or real-world data are highly desirable. The project is part of the Graduate School studentships competition for October 2026 entry, with funding available for successful applicants.

For more information and application instructions, visit the Kingston University PhD Studentships page and the Faculty of Engineering, Computing and the Environment research webpage. The application deadline is March 4, 2026.