Dynamic Resource Allocation for AI/ML Applications in Edge Computing: Framework Architecture and Optimization Methods

Abstract

This scholarly paper introduces an extensive architectural framework and optimization strategies designed specifically for dynamic resource allocation in edge computing environments, with a focus on AI/ML applications. The rise of edge computing presents a viable solution for managing the computational complexities of AI/ML tasks by utilizing resources in proximity to data sources. Nevertheless, effective resource allocation encounters significant hurdles due to the diverse and ever-changing nature of edge environments. In addressing these challenges, the paper introduces an innovative framework that integrates dynamic resource allocation methodologies with the unique requirements of AI/ML applications. This framework encompasses a range of optimization techniques customized to efficiently distribute resources, taking into account factors such as workload attributes, resource availability, and latency limitations. Through extensive simulations and evaluations, the study showcases the effectiveness of the proposed approach in enhancing resource utilization, reducing latency, and bolstering overall performance for AI/ML workloads within edge computing scenarios.