Task Offloading And Caching Based On Mobile Edge Computing

Mobile edge computing is an emerging architecture that enhances the power of mobile cloud computing by deploying cloud resources,such as storage and computing ability,to the edge of wireless access networks.This provides users with powerful and efficient computing services,storage services,energy efficiency,mobility,location,and context-aware support.Mobile edge computing supports a variety of innovative applications and services that require ultra-low latency.However,in the research on the mobile edge computing task offloading,the computing task offloading by high-speed moving users will lead to frequent switching of computing tasks in the system,which will introduce additional transmission delay and reduce user experience.At the same time,when the mobile node is used as the mobile edge cache server,the mobility and the unknown popularity of the cache content will affect the existing cache strategy and reduce the hit rate of the cache.In addition,user mobility makes the interaction with mobile edge computing server more frequent,which leads to the increase of energy consumption of user equipment.This thesis focuses on the above challenges,mainly in the following three aspects:Aiming at the problems of reduced throughput and increased energy consumption caused by vehicle users offloading computing tasks to mobile edge computing servers,a new offloading paradigm for subsequent computing tasks has been developed in which a mobile edge computing server can be deployed on a bus to provide additional computing resources.On this basis,in order to solve the problem of which mobile edge computing server the vehicle user offloads the computing task to,a deep Q-learning computing task offloading scheme with a priori knowledge is proposed.The simulation results show that the proposed deep Q-learning task offloading scheme with prior knowledge has higher efficiency and performance advantages.In order to solve the mobility problem caused by the deployment of the mobile edge computing cache server on the bus and the unknown popularity of the cacheable content,this thesis proposes a mobile edge computing cache strategy based on Multi-arm Bandit.First,each bus cache server is treated as an agent,constantly learning content requests from nearby users and updating the cached content in the limited cache space.On this basis,combined with the characteristics of popular content,the learning efficiency of bus cache server is accelerated and the learning effect is improved.The simulation results show that the proposed strategy can not only improve the cache hit ratio of the bus cache server,but also better adapt to the dynamic changing environment.Compared with the traditional strategy,the cache hit ratio can be increased by more than 10%.Aiming at the problem that the user's mobility makes it interact with the mobile edge computing server frequently,which leads to the increase of energy consumption and shorter endurance of the user equipment,this thesis proposes a low-energy user equipment receiving strategy in the mobile edge computing scene based on the discontinuous reception.This solution can set different modes according to different traffic.Each mode has a different discontinuous reception configuration.The base station sends a power saving signal to the user equipment to realize the conversion of the user equipment in different modes and save energy for the user equipment.On this basis,for the physical downlink control channel monitoring in the discontinuous reception,the user equipment can report the user equipment auxiliary information to the base station according to the channel conditions.The auxiliary information includes the physical downlink control channel aggregation level preferred by the user equipment.After receiving the information,the physical downlink control channel information can be sent to the user equipment at the corresponding aggregation level,and the user equipment performs the blind decoding of the physical downlink control channel under the aggregation level,which reduces the number of blind decoding of the physical downlink control channel and achieves the further improvement of the user equipment.The simulation results prove the effectiveness of the user equipment power saving solution when the user equipment receives content.