Research On Task Processing Strategy Of Internet Of Things In Edge Computing

The 5G network puts forward lower latency requirements for tasks,making the original network architecture incompetent.Mobile Edge Computing(MEC)is proposed and considered to be a feasible architecture to solve network delay requirements.This architecture arranges computing resources on the base station side.Compared with cloud computing,this architecture achieves closer services and satisfies terminal computing resources.The caching of popular resources is a kind of thought to reduce the redundancy of communication data.The introduction of caching in MEC can preset the resources that users need to upload in advance,thereby reducing the cost of task uploading and further improving network performance.The terminal offloads its own computing tasks to the MEC server,which can reduce task delay and save terminal energy consumption when its own resources are limited.In a real network,resources such as computing and communication are often requested by multiple tasks at the same time.How to choose an offloading strategy and a resource allocation strategy has a critical impact on the performance of the entire network.The size of the cache introduced in MEC is limited,and the cost function reduction brought by different cache schemes has significant differences.Therefore,the main research of this thesis is as follows.An offloading strategy and resource allocation scheme are proposed for multiple users requesting tasks at the same time under the coverage of a single non-cached MEC server.In this scenario,the MEC server is not only requested by the new task,but its own old task is still being calculated.The cost function of a task is a weighted summation of delay and energy consumption,and the weight of delay energy for each task is different.Consider including the resources occupied by running tasks in the server into the scope of optimization to reduce the cost of the entire system.The unloading strategy includes whether the new task is uploaded or not and the resources allocated,and whether the running task is suspended to free some resources.The objective function of this problem is the NP problem.In order to solve this problem,this thesis uses an improved whale algorithm to optimize it.The simulation results show that the proposed strategy can significantly reduce the cost of the system.Introducing cache resources into the MEC server is an effective solution to improve service performance.In order to take advantage of the MEC server with enhanced cache,it is necessary to reasonably preset user resources on the base station side in advance.Each task itself has three basic attributes:popularity,space size,and required calculations.The optimization goal of the cache is to increase the hit rate and the benefits of the hit itself.The determination of the cache solution in the network and the initiation of the computing task can be divided into two actions.When solving the cost optimization function in this scenario,it is also divided into two steps:cache solution design and computing task optimization.First,the caching scheme is transformed into a knapsack problem,and the storage scheme is obtained recursively using dynamic programming.Then,on the basis of the storage scheme,the whale algorithm is used to optimize the calculation task to minimize the cost.The simulation results show that Under the condition of the same optimization scheme for communication and computing resources,a reasonable cache placement strategy can maximize the benefit brought by cache.Compared with the algorithm benefits and hardware benefits brought by using optimization scheme without cache and random allocation of resources with cache,the advantages will change under different parameters.On the basis of the previous two parts of research,the problem is extended to the scenario of multi-MEC cooperative buffering and computing.In this scenario,each base station has certain computing and storage resources,and the user density in different cells is different.Multiple MECs cooperate in caching and computing resources to jointly serve all users.In this chapter,the problem is decomposed into cache placement,offloading strategies,and resource allocation.A greedy algorithm that comprehensively considers file popularity and richness is designed to solve the storage problem,and particle swarm optimization is used to optimize offloading decision-making and resource allocation.The simulation proves the effectiveness of the algorithm and the gains from collaboration.