DESIGN OF AN EFFICIENT BIOINSPIRED CONSTRAINT ENFORCEMENT SCHEDULING MODEL FOR FOG DEPLOYMENTS UNDER HETEROGENEOUS TRAFFIC SCENARIOS

Abstract

In order to optimize efficiency of fog environments, task planning is essential for efficiently scheduling tasks, meeting deadlines, and maximizing resource utilization. This paper presents a novel and efficient bioinspired constraint scheduling model for fog deployments under heterogeneous traffic scenarios. The proposed method incorporates bio-inspired optimization techniques and a dependency-aware clustering model to improve task-scheduling effectiveness. Necessity of this research arises due to increasing demand for efficient task planning techniques in cloud computing environments. Efficient task planning requires taking into account task requirements, VM capacity, dependencies, and deadlines, while ensuring a high deadline hit ratio, reduced cloud effort, and task diversity. To address these challenges, we propose a comprehensive approach that optimizes task scheduling using bioinspired optimization techniques. The crux of our strategy is the development of a dependency-aware clustering model that groups tasks based on a scoring metric that takes task completion time and deadlines into account. We employ an efficient fusion of kMeans, and Hierarchical Clustering techniques to enable clustering of tasks. Consequently, we use the Grey Wolf Optimization (GWO) model to map VM configurations to the corresponding workloads. Taking into account task capacities and deadlines, the GWO model optimizes both task clusters and VM configurations, ensuring efficient scheduling operations. Utilizing task-level & VM-based constraints enforcement during scheduling operations is a significant advantage of our method. By incorporating these constraints, we are able to effectively resolve dependency issues and enhance the overall performance of the scheduling process. In addition, the proposed method incorporates a fitness function that takes into account both task deadlines and capacities, thereby enhancing the mapping process and resulting in enhanced scheduling outcomes. To assess the efficacy of our proposed method, we compare it to existing scheduling models. Our model achieves a 3.5% better deadline hit ratio, 2.9% greater scheduling efficiency, 4.9% greater task variety, and 3.2% less computing effort than current scheduling models under identical scheduling scenarios, as demonstrated by the results.

Keywords: Cloud Computing, Task Planning, Bioinspired Optimization, Constraint Enforcement, Fog Deployments