COMPARATIVE ANALYSIS OF HEURISTIC AND AI-BASED TASK SCHEDULING ALGORITHMS IN FOG COMPUTING: EVALUATING LATENCY, ENERGY EFFICIENCY, AND SCALABILITY IN DYNAMIC, HETEROGENEOUS ENVIRONMENTS
Kaushik Sathupadi
Staff Engineer, Google LLC, Sunnyvale, CA
https://orcid.org/0009-0007-1189-2293
Keywords: AI-based Scheduling, Fog Computing, Heuristic Algorithms, Resource Utilization, Scalability, Task Scheduling
Abstract
Fog computing extends cloud capabilities to the edge of the network, offering a distributed infrastructure to handle latency-sensitive, bandwidth-intensive applications. Task scheduling process in fog computing determines how computational tasks are allocated to heterogeneous, resource-constrained fog nodes. The dynamic and unpredictable nature of fog environments, characterized by stochastic task arrivals and fluctuating resource availability, adds complexity to the scheduling process. This paper presents a comparative study of heuristic and AI-based scheduling algorithms, focusing on their effectiveness in managing task allocation under dynamic and heterogeneous conditions. Heuristic algorithms are known for their computational efficiency and low complexity, making them suitable for resource-limited environments. However, they often lack the adaptability required to handle the stochasticity of real-world fog networks. AI-based scheduling algorithms using machine learning and optimization techniques can provide flexibility and adaptability by learning from system dynamics and predicting future states. This study evaluates these approaches using performance metrics such as latency, energy consumption, resource utilization, and scalability. The findings reveal trade-offs between the computational overhead associated with AI-based methods and their superior performance in dynamic, heterogeneous environments.
