Research On Resource Management Mechanism Based On Multi-Armed Bandit In Edge Computing

The ubiquity of smart devices has propelled the popularity of smart applications due to the continuous development of the Internet of Things.As these applications become more functional,resource-constrained mobile smart devices are increasingly demanding computing.Cloud computing has centralized configurable and powerful computing resources,making it easily accessible to mobile users,but due to various unpredictable network conditions and the sensitivity of emerging applications to latency,transporting all data to a congested backbone network changes is not feasible.As a solution,Mobile Edge Computing(MEC)emerges as the times require,as a new computing paradigm to achieve service provisioning near smart devices,so that analysis and knowledge generation occur closer to the data source,And provide delayed response services.However,the massive amount of private data involved in the network edge layer brings privacy security and delay requirements.In reality,the resources of the user end are limited,and the tasks that can be processed are also limited.This thesis studies the problems involved in resource management in mobile edge computing,and the work is as follows:(1)In this thesis,we propose a new resource management mechanism to optimize the service latency of users while considering privacy protection.We transform the dynamic resource management problem into a multi-armed bandit(MAB)problem,and give a new method to quantify privacy,and propose a privacy-preserving awareness resource allocation based on a multi-armed bandit algorithm.Through theoretical analysis and simulation experiments,it is verified that the algorithm can not only achieve the purpose of privacy protection,but also control the device deadlock rate to zero,which can achieve good optimization performance.At the same time,we prove the convergence of the proposed algorithm.(2)In this thesis,we combine the ideas of multi-armed slot machines and knapsacks,and propose a static batching algorithm based on Bwk(Bandits with Knapsacks)to maximize the use of limited resources to process as many tasks as possible.Theoretical and simulation results show that the proposed scheme can achieve the goals of saving resources and increasing the number of completed tasks.