Cooperative Resource Allocation In Fog Computing

With the wide use of the Internet in social production and life,the number of devices accessing the network is increasing rapidly,and the amount of data is increasing dramatically,and there are more and more computing tasks with low delay,high reliability and large calculation quantity requirements.The traditional computing method is to send computing tasks to the cloud data center for processing,which will lead to network congestion,and the high delay caused by the long transmission distance between the data center and the end user,also cannot meet the needs of real-time applications.Fog computing is a new computing mode.Compared with the centralized computing resources in cloud data center,fog computing distributes computing resources to network edge devices,so that user data can be processed at the edge of the network.Fog computing can effectively reduce the computing delay by processing most of the user data at the edge of the network,which supports new delay sensitive applications and solves the problem of service demand with low delay.This computing mode also reduces the amount of data uploaded to the core network and effectively avoids network congestion.Each fog node in fog computing is a device with limited computing resources and storage resources,so it is difficult for these devices to allocate resources to meet the needs of users through cooperation.At the same time,the diversity of computing and storage resource requirements of user computing task makes resource allocation more complex.So how to allocate the limited resources in fog computing to deal with a large number of user data is very challenging.This thesis studies the cooperative resource allocation of fog computing,and the details are as follows:In this thesis,aiming at the scenario of cooperative task processing between fog nodes,a nonconvex optimization model with the minimum average task delay as the goal is established.According to the nonconvex model,a lower bound problem is designed,which is a convex optimization problem.By iterating the lower bound problem repeatedly,the approximate solution of the original problem is obtained.In addition,the convergence of the algorithm is proved,and the error between the lower bound problem and the original problem is analyzed.Aiming at the scenario of cooperative task processing among users,fog nodes and cloud,this thesis also establishes a nonconvex optimization model with the maximum weighted access as the goal.According to the nonconvex model,the lower bound problem and the upper bound problem are designed.Both of them are convex optimization problems.By iterating the lower bound problem and the upper bound problem repeatedly,the approximate solution of the original problem is obtained.In addition,the convergence of the algorithm is proved,and the error between the upper bound problem and the original problem is analyzed.Based on the above two scenarios,the simulation results show that the resource allocation decisions and task offloading decisions based on the two algorithms can achieve the system optimization respectively.And the decision meets the requirements of task processing.The device can allocate the corresponding resources for task processing.The simulation results show that the more fog nodes cooperate in the network,the less time each task consumes on average.The simulation results also show that with the increase of task arrival rate,the access rate of the system will gradually decrease,and the tasks that consume more system resources will be partially accessed first.