Research On Mobile Edge Computing Task Offloading And Resource Allocation Management

With the rapid development of new services such as high-definition mobile video,AR/VR,car networking,and intelligent manufacturing,the network is required to provide specific capabilities such as ultra-low latency,high speed,and customization.These requirements pose severe challenges to the next generation mobile communication network represented by 5G.Mobile Edge Computing（MEC）system has abundant computing and storage resources,and has the characteristics of proximity access.It allows mobile terminals to offload tasks to the edge cloud,thereby reducing task processing delay and terminal's energy consumption.In order to solve the limitation of mobile terminal resources and improve task processing efficiency,this dissertation proposes a joint multi-user and multi-MEC task offloading and resource allocation joint optimization algorithm and a Q-learning-based end-edge-cloud collaboration fast offloading algorithm,and through simulation,the proposed algorithms are analyzed and verified.The main works of this dissertation are as follows:（1）An edge-end collaboration task offloading and resource allocation algorithm based on multi-user and multi-MEC scenarios is proposed.With the goal of maximizing the total benefit of tasks,the optimal task offloading resource scheduling problem based on Lyapunov theory is formed by the constraints of service Qo S guarantee and resource limitation.Since the problem is NP-hard,it is decoupled into a channel resource allocation problem solved by KKT condition and a 0-1integer programming problem about task assignment.Compared with the traditional algorithm,the complexity of the proposed algorithm is ^{O（7）mn（8）},and the efficiency is improved by about 20%,and the delay is reduced more than 15%.（2）Aiming at the fast-moving scenario of mobile terminal,a fast offloading algorithm based on Q-learning is proposed,and the task offloading system model in the fast-moving scenario of terminal is elaborated.The task offloading algorithm first divides the task into multiple subtasks based on the number of MEC servers and the remaining available resources of the MEC server.Considering the two factors of delay and energy consumption comprehensively,with the goal of maximizing the benefits of the user terminal,the TOPSIS algorithm is used to obtain the immediate reward obtained by the user terminal after making the offloading decision.Considering the randomness of the assignable communication resources,this problem is further formulated as a semi-Markov process.Finally,the Q-learning algorithm is used to update the cumulative discount rewards by combining instant rewards and experience rewards to make the best offloading decisions and resource allocation strategies to obtain the maximum user terminal benefit.Then the overall flow of the fast offloading algorithm based on Q-learning is described in detail,and the time complexity of the algorithm is analyzed.Finally,the proposed algorithm is compared with other algorithms through simulation.The simulation results show that compared with other algorithms,the proposed algorithm has faster convergence speed,smaller average delay of offloading,and a lower user terminal energy consumption.