Edge container resource allocation method based on deep reinforcement learning

Abstract

The invention discloses an edge container resource allocation method based on deep reinforcement learning, which comprises an Actor network, a Critic network and an echo state network (ESN), and effectively solves the problem of container resource allocation of an edge computing environment of time delay sensitivity application. According to the resource allocation method provided by the invention, an end-to-end time delay model is established on the basis of an M / D / 1 queuing model to obtain end-to-end time delay of a service flow s data packet on a container n, and a deep reinforcement learning model is adopted to solve the problem of edge container resource allocation. An ECA3C network is obtained by improving a traditional A3C algorithm through an echo state network (ESN), a resource allocation strategy At is allocated to different container clusters Sz and t in a resource pool z and used for solving the problem of container resource allocation in an edge computing environment, andthe ECA3C network changes an end-to-end delay reward value rt to obtain an end-to-end delay reward value rt. Therefore, the method is suitable for various edge computing environments.

Description

technical field [0001] The invention relates to the field of resource allocation, in particular to an edge container resource allocation method based on deep reinforcement learning. Background technique [0002] With the development of science and technology, the types and quantities of service request streams of various applications have exploded. The resource pool needs to be quickly destroyed and deployed at different times without affecting the quality of service. It is completely suitable for different container services. The combined strategy and lightweight container technology allow us to reallocate the resources of the resource pool in a short period of time, and minimize the impact on the service quality of service requests during the resource allocation process, which enhances the deployment of applications flexibility and efficiency. When an application processes multiple containers, it needs a series of containers and network links connecting the containers, wh...

AI Technical Summary

Problems solved by technology

[0005] In view of the deficiencies in the existing technology, the purpose of the present invention is to provide a method for allocating edge container resources based on deep reinforcement learning, which effectively solves the problem of container resource allocation for the edge computing environment of delay-sensitive applications

Images