Adaptive Scheduling Based on Intelligent Agents in Edge-Cloud Computing Environments

Abstract

Scheduling in cloud computing environments has been extended to support the Internet of Things (IoT) applications, which require additional quality of services such as energy consumption and real-time properties. To this end, edge-cloud computing environments are prevalently deployed by encompassing the fog management layer. However, traditional scheduling techniques for cloud tasks have limited capabilities to support real-time properties required for IoT applications. In this paper, we propose a deep learning-based dynamic cloud scheduling technique using intelligent agents, which intelligently adapt to users’ requirements and selective quality of services based on distributed learning in edge-cloud computing environments. The proposed cloud task scheduling method is composed of two logical components: distributed learning management (learning distribution and aggregation) and intelligence management of multi-agents, which are independent of each other. The performance results show that the self-employed agents intelligently adapt to their environments and perform hyperparameter learning for efficient and effective task scheduling in edge-cloud computing environments.