Research On Computation Offloading And Load Balancing Algorithm In Mobile Edge Computing Networks

With the rapid development of the mobile Internet and the Internet of Things,more and more terminal devices and new applications(such as Augmented Reality,face recognition and interactive games,etc.)have appeared in people's daily lives.These new applications usually have the characteristics of being computationally intensive and delay sensitive,and have higher requirements for the computing and storage capabilities of terminal equipment.The advent of Mobile Edge Computing(MEC)has emerged as a promising solution to the problems of insufficient computing and storage capacity of terminal devices.User equipment can offload complex computing tasks that are difficult to handle to the MEC server at the edge of the mobile network,and use their rich computing and storage resources to perform task processing,effectively improving user service quality.While MEC brings many advantages,there are many technical challenges such as how to formulate a reasonable and efficient computation offload mechanism based on limited computing resources(for task processing)and wireless resources(for task transmission),and the strategies of distributing the offload tasks in the MEC servers caused load balancing problems.This thesis focuses on computational offloading and load balancing algorithms in MEC networks.The specific content is as follows:1.Aiming at solving the computational offloading problem in the single-cell multiuser MEC scenario,we firstly considered the user offloading decision and the impact of limited wireless and computing resources on computational offloading,and a joint optimization problem of tasking offloading decision and resource allocation was formulated.First,an adaptive genetic algorithm is applied to make the offloading decision and the subsequent update operations.The original problem is decomposed into two subproblems,namely power allocation and computing resource allocation,following each offloading decision update.Then,according to the theory of convex optimization and quasi-convex optimization,we employed the binary search method and the Lagrangian multiplier method to obtain the optimal solutions of power allocation and computing resource allocation,respectively.Simulation results show that the proposed scheme reduces the total user overhead while ensuring the user delay constraints,and effectively improves system performance and user experience quality.2.In order to deal with the problem of computational offloading in a dense heterogeneous network MEC scenario,we considered the impact of co-channel interference in different smallcells on the performance of computational offloading,and a joint optimization problem of offloading decisions and resource allocation was formulated.Firstly,the chaotic mutation binary particle swarm algorithm was used to optimize the user's offloading decision.Under the specific offloading decision,the Lagrangian multiplier method was used to allocate computing resources to the user.Under the condition,the improved Kuhn-Munkre algorithm was used to perform subchannel allocation for offloaded users.Simulation results show that the proposed scheme can save more overhead and effectively improve system performance than other schemes.3.As for the problem of load balancing of MEC servers in the future,we considered the changes in the amount of offload tasks over the time in the network,the MEC server where a very low amount of tasks to cope with was necessary to hibernate to save energy consumption.First,the M/M/m multi-service queuing theory was used to model the amount of offload tasks in the network,and then based on the amount of offload tasks in the network,a set mean iterative comparison algorithm was used to filter out the MEC server set with less offload tasks.The MEC servers in the collection are judged one by one and the hibernation operation is performed.Simulation results show that the proposed scheme can significantly reduce the energy consumption of the system.