Design a Hierarchical Resource Allocating Approach by Using Dual Q-Learning in Deep Reinforcement Algorithm

With the advent of the Internet of Things (IoT), the number of devices that work in these environments has greatly increased, which has caused major changes in the structure of data processing and information storage. In this regard, cloud computing is an Internet-based computing platform that provides the necessary processing resources for users. However, due to the ever-increasing volume, speed and communication technologies, the current model of cloud processing can hardly provide satisfactory performance quality. So, proper management of the incoming requests is very critical for the continuous operation. In this project, in order to increase the speed of resource allocation operation, virtual machines are ranked by game theory, which significantly reduces the time and computational complexity of the whole process. The combination of these three approaches and their integration in sequence increases the speed of allocation performance and reduces the costs associated with it. In this article, two characteristics of activity completion time and service quality have been evaluated. According to the obtained results, applying the proposed structure on the Borg dataset provided by Google has yielded results in the range of more than 35% for reduction in CPU usage cost compared to other existing methods.