Mobile edge network intelligent resource allocation method capable of dividing tasks

Abstract

The invention relates to a mobile edge network intelligent resource allocation method capable of dividing tasks, which comprises the following steps of: dividing serial tasks generated by a terminal to obtain a plurality of subtasks, and establishing an unloading task model; respectively establishing a time delay model and an energy consumption model for the subtasks according to a local execution mode or an unloading execution mode, and defining an unloading joint target optimization function based on a multi-user serial dependent task; in a multi-server scene, establishing a Markov game model according to a cooperative competition relationship of multiple users to wireless communication and computing resources, and optimizing the unloading joint target optimization function; and in a time-varying environment, taking each terminal as an independent agent based on part of system state information to execute a reinforcement learning algorithm to solve the Markov game model, and determining an unloading strategy, sub-channel selection, transmitting power and resource allocation quantity. According to the mobile edge network intelligent resource allocation method, server resources can be reasonably allocated, fragmented resources can be fully used, the terminal user experience is ensured, and the network operation stability is improved.

Description

technical field [0001] The present invention relates to the technical field of edge computing and artificial intelligence, in particular to a mobile edge network intelligent resource allocation method that can divide tasks. Background technique [0002] With the continuous development of communication technology, a large number of emerging Internet interactive applications have emerged. These applications have continuously increased requirements for data transmission, computing power and delay of mobile devices, so they are not suitable for applications with poor computing power and limited battery capacity. Execute on smart devices. In addition, a single cloud architecture requires long-distance data transmission, which is difficult to meet the requirements of low latency and large bandwidth on the terminal side of ultra-dense wireless networks under the next-generation communication framework. For this reason, mobile edge computing technology, as a specific implementation...

AI Technical Summary

Problems solved by technology

However, this method is oriented to single-server offloading scenarios, and the problem solving dimension is high, which cannot meet the needs of multiple terminals in a dense network, and the algorithm has poor scalability.

[0004] The deficiencies of existing technologies are mainly reflected in four aspects. First, the scenario is too simple. Most researches are oriented to single / multi-terminal single-server scenarios, taking into account the competition of computing and communication resources between devices, but ignoring the selection of offloading servers, inter-server Problems such as load balancing, resource scheduling and allocation; second, unloading tasks cannot be divided. Existing research is limited to 0-1 unloading of indivisible atomic tasks, ignoring the potential parallelism among divisible tasks, and unable to effectively utilize server fragmentation resource

The third is that the optimization goal is too single, only considering the two aspects of delay and energy consumption, ignoring other factors that affect system performance, such as tasks with different urgency levels need to be handled differently; finally, the central offloading strategy is less adaptable to dynamic environments. Poor, make a unified decision based on the collected global information, the central control node needs to bear huge calculation and traffic pressure, and it is easy to become the bottleneck of the entire system

Images