Optimizing energy and latency in edge computing through a Boltzmann driven Bayesian framework for adaptive resource scheduling.

This paper presents a new approach based on Boltzmann Distribution and Bayesian Optimization to solve the energy-efficient resource allocation in edge computing. It employs Bayesian Optimization to optimize the parameters iteratively for the minimum energy consumption and latency. Coupled with this, a Boltzmann-driven probabilistic action selection mechanism enhances adaptability in selecting low-energy tasks by balancing exploration and exploitation through a dynamically adjusted temperature parameter. Simulation analysis demonstrates that the new method can decrease energy consumption and average delay much lower than Round-Robin and threshold-based algorithms. The feature of temperature adaptation within Boltzmann further guarantees the achievement of the optimal scheduling actions while ensuring flexibility in the case or altering load percentages. Cumulative energy savings varied up to 25% compared to baseline methods, demonstrating the applicability of the framework in real-time, energy-aware applications at the edge. This work demonstrates the viability of combining probabilistic selection with parameter optimization, setting a new benchmark for energy-efficient resource scheduling. Such findings create possibilities in expanding the existing literature on the use of hybrid optimization methods to enhance sustainable computing solutions in the context of distribution systems.