Research On Multi-access Edge Computing Task Offloading Supported By MDMA

In order to enable mobile devices to process data faster and more energy-efficient,5G proposes Multi-access Edge Computing(MEC)technology.The MEC deploys the server near mobile devices and use the multi-access technology to build communication links which make mobile devices quickly offload partial data to the server,and then both devices and the server calculates all data simultaneously.Therefore,MEC is closely related to multi-access technology and will be updated and adapted with the development of the latter.At present,the vortex electromagnetic wave and the corresponding Mode Division Multiple Access(MDMA)technology are included in 6G planning.The MEC needs to be combined with the MDMA for better performance.Therefore,the research questions of the thesis are how the MDMA is integrated into the MEC and what the performance of the MDMA-MEC is.The MDMA-MEC research includes offloading decision and resource allocation.The offloading decision is the ratio of data uploaded by devices,and resource allocation is the scheme for allocating computing and communication resources to all devices.The thesis studies the offloading decision and resource allocation of the MDMA-MEC for minimizing average latency,minimizing average energy consumption and joint optimization both of them:(1)In order to obtain the minimum average latency of the MDMA-MEC,the thesis combines parameters,which are the mode of vortex electromagnetic wave and the frequency into the resource block,and adjusts the MDMA channel capacity formula based on it.Then the MDMA-MEC average latency formula is formed by combining the MDMA channel capacity formula with the MEC latency formula.Because the MDMA-MEC average latency formula has the characteristics of different variable types and large infeasible area,the Modified Differential Evolution(MDE)and the Modified Tournament Selection(MTS)are proposed to respectively process them.The simulation results show that the MDMA-MEC can obtain smaller average latency than MEC without MDMA,and hybrid algorithm can solve the problem of minimizing average latency more effectively.(2)In order to obtain the minimum average energy consumption of the MDMA-MEC,the thesis adds the MDMA channel capacity formula to the MEC energy consumption formula to construct the average energy consumption formula of the MDMA-MEC.The formula has the characteristics of relatively few variables,and the values are mostly 0 or 1.Therefore,the Modified Genetic Algorithm(MGA)is used,and crossover and mutation operations are formulated for it.At the same time,the Water Filling(WF)is used to obtain the maximum channel capacity of MDMA to assist the MGA.The simulation results show that the MDMA-MEC can obtain a smaller energy consumption value than the current MEC,and the relationship between the minimum average energy consumption value and the relevant parameters.The proposed algorithm can also effectively obtain a smaller average energy consumption.(3)In order to jointly optimize latency and energy consumption,the thesis takes into account the bias between latency and energy consumption of the MDMA-MEC and the benefit,and constructs the average biased benefit formula by weighting the ratio between the actual value and the limit value of latency and energy consumption.The latency ratio and energy consumption ratio are related to the minimum average latency and minimum average energy consumption in the previous chapters.Therefore,the algorithm that minimizes average latency ratio and minimizes average energy consumption ratio in turn is proposed to find the minimum value,and its core is built by integrating the previous algorithms.Simulation results show that the MDMA-MEC can obtain a smaller average biased benefit than the current MEC,and the characteristics related to the number of modes and devices are similar to the previous simulation results.The results show that the average latency ratio is negatively correlated with the bias coefficient,while the average energy consumption ratio is opposite.The algorithm can effectively solve the problem of minimizing the average biased benefit.