Energy-Efficient Edge Computing Architectures for AI Workloads: A Comparative Analysis in Cloud-Driven Environments

In the landscape of cloud-driven environments, the convergence of artificial intelligence (AI) workloads with edge computing architectures holds promise for optimizing computational efficiency and minimizing latency. This paper conducts a comprehensive comparative analysis of energy-efficient edge computing architectures tailored specifically for AI workloads. Beginning with an overview of the escalating importance of edge computing within the realm of AI-driven applications, we underscore the necessity for architectures that effectively balance computational prowess with energy conservation. Subsequently, we scrutinize various existing paradigms of edge computing, including Infrastructure as a Service (IaaS), Platform as a Service (PaaS), and more, assessing their suitability for accommodating AI workloads and delving into the intricate interplay between performance, energy consumption, and scalability. Our analysis encompasses a diverse array of architectural approaches, spanning fog computing, mobile edge computing (MEC), and distributed edge computing, and scrutinizes their efficacy in facilitating AI inference tasks across a spectrum of deployment scenarios. Another thing we look at is ways to make edge settings run faster and use less energy, like task sharing, resource allocation, and job timing. AI jobs can be done with edge computing that doesn't use a lot of power. Through in-depth case studies and a full review of the empirical literature, we show the pros and cons of each design model. This study not only helps people make better decisions, but it also moves cloud-based possibilities forward.