Research On Optimization Method Of Computation Offloading For Fog Wireless Access Network Based On Deep Learning

Fog wireless access network computing offload is a new type of network technology that offloads the local tasks of end users to service nodes in the fog wireless access network for processing,so as to alleviate the insufficient local resources of end users to meet the needs of computing tasks.While establishing a high-speed and stable network connection through the fog wireless network,users can also directly obtain data storage,computing and processing services from fog nodes and cloud nodes in the network.In order to fully tap the utilization potential of fog wireless access network resources and improve the user experience of users in the network,establishing a reasonable and efficient calculation shunt decision mechanism is the key and challenge.Based on the above-mentioned technical background,this article mainly researches the computing offloading technology in the fog wireless access network.The main research work points are as follows:1.In the cloud and fog collaboration scenario,the shunting decision and resource allocation optimization methods when multi-users perform computing task shunt processing are studied.By establishing a computing shunt system model for cloud and fog collaborative processing,under the premise of limited shunt system resources,an optimization problem with the goal of minimizing system task processing energy consumption is summarized.The shunting decision problem of computing tasks is a nonlinear and non-convex integer programming problem.When there are many users in the system,traditional optimization algorithms are not applicable due to the high complexity of the algorithm.The deep reinforcement learning technology can establish the mapping relationship between the user state in the system and the optimal shunting decision solution through the neural network,so as to avoid the difficult problem of solving the integer programming problem.Therefore,this paper studies a joint optimization method of computing shunting and resource allocation based on deep reinforcement learning,using the powerful and effective decision-making ability of deep reinforcement learning technology to optimize the shunting decision and resource allocation of the computing shunt system,and effectively improve Improve the operating efficiency of the system.2.Researched the optimization problems of computation offloadting and shunting decision-making and resource allocation for multi-task concurrent offload processing in real-time business scenarios.The first thing to consider is that each user will have multiple tasks that need to be analyzed at the same time for each diversion decision.The delay model of the system needs to consider the order of entering the task queue between different tasks.Secondly,the sub-carrier resources of the fog node are limited.When there are many tasks for offloading processing at the same time,reasonable allocation of the sub-carrier resources will be another key factor affecting the quality of system service.By using the advantages of the deep deterministic strategy gradient algorithm in terms of learning efficiency and convergence speed,it can effectively solve the problem of high algorithm complexity caused by offloading decision and subcarrier allocation problems.At the same time,in order to prevent the algorithm from slowing down due to excessive use of neural networks,the system's resource allocation problem can be solved by traditional convex optimization schemes.The final simulation experiment proved that the designed offloading decision-making and resource allocation optimization algorithm based on the depth-determined strategy gradient of multi-task concurrent shunt processing can effectively reduce the system performance overhead and improve the system's service quality.This article mainly focuses on the problem of poor timeliness in computation offloading and resource allocation in fog wireless access networks in real-time business scenarios.After studying the efficiency and effectiveness of deep reinforcement learning in decision-making problems,the technology is introduced to optimize the computation offloading and resource allocation of the fog wireless access network,so as to optimize the resource allocation of the fog wireless access network.The purpose of improving the efficiency and service quality of the fog wireless access network system.