Resource Management Technology For Mobile Edge Computing In Heterogeneous Network

With the rapid development of modern communication technologies and the gradual diversification of application scenarios,the demand for computation-intensive and latency-sensitive tasks from users is exploding.Mobile Edge Computing(MEC)is considered to be one of the key technologies to solve current needs.In the future,the MEC network will develop toward an intelligent heterogeneous network with cloud-edge-end collaborative operation.When there are multiple types of computing entities,how to achieve joint optimization of heterogeneous communication,computing,and caching resource allocation through collaboration between different nodes to reduce operating energy consumption and ensure end-to-end latency is a major research hotspot for MEC network to be solved urgently.Considering the above issues,this thesis will conduct research on MEC resource management technology in heterogeneous network from the following two aspects.For the static MEC system,an MEC heterogeneous network that supports device to device(D2D)communication is constructed to improve the computing performance of the system.The network includes both D2 D offloading and traditional MEC offloading.The task execution cost is modeled to weigh the impact of latency and energy consumption on the network performance,with the goal of minimizing the task computation cost for all user devices.Based on this,a joint resource allocation and computation offloading algorithm is proposed in this thesis,which first finds the optimal resource allocation by solving the Lagrangian equation under the Karush-Kuhn-Tucker(KKT)condition for a fixed computational offloading pair.Then the sum execution cost is minimized by selecting the offloading pair based on the first step.Simulation results demonstrate the effectiveness of the algorithm,which can significantly reduce the system execution cost.For Edge Intelligence(EI),this thesis proposes a joint communication,computing and caching optimization strategy for EI,which integrates and jointly optimizes the collaborative caching model and task offloading model of edge nodes to improve the quality of service of users.First,a Co-cache-assisted computing algorithm in heterogeneous network is proposed with the goal of minimizing the cost of performing tasks for all user devices by jointly optimizing caching decisions,user-helper connections,and resource allocation through collaborative content delivery from several helpers to users.Furthermore,a deep reinforcement learning algorithm is proposed to solve the optimization problem in order to adapt to the dynamic task request and environment change.Simulation results show that the algorithm can converge effectively and demonstrate that the Co-cache-assisted computing algorithm can significantly reduce the task execution cost.