Research On Mobile Edge Computing Network Nodes Selection Strategy

Mobile edge computing(MEC)technology,as one of the core technologies of the 5th generation mobile communication system,sinks cloud computing capabilities and business platforms to the edge of the network(base station / terminal).The wireless access network can provide cloud computing and IT services for computing-intensive services and delay-sensitive services to users nearby,such as car navigation and virtual reality technology.With the base station / terminal deploying the MEC server,it will affect the scheduling of communication,computing and cache resources of traditional networks.Therefore,this article conducts research and analysis on nodes selection strategies for the relay communication and D2D(Device-to-device)traditional communication networks with MEC.Combining MEC and relay technology,MEC servers are deployed on the relay nodes side.This paper studies the influence of relay computing capabilities on the relay nodes selection strategies.Based on the single-user multi-relay selection model,this paper proposes three relay nodes selection strategies : communication-only relay selection(CORS),computing-only relay selection(CPORS)and end-to-end latency-best relay selection(LBRS)that comprehensively considers communication and computing.Because the traditional rate outage probability is often used to analyze the behavior of the physical layer of the cooperative communication system,it cannot intuitively show the influence of computing power on the selection of relay nodes.Therefore,the end-to-end delay outage probability is used as the evaluation criterion of system performance.The delay outage probability and diversity order of the three strategies are derived,respectively.Through simulation,it is found that the CPORS strategy is superior to the CORS strategy under low signal-to-noise ratio.The influence of relay computing capabilities on the link transmission delay is gradually reduced under high signal-to-noise ratio and the CORS strategy is superior to the CPORS strategy.The theoretical formula shows that under the condition of extreme high signal-to-noise ratio,the delay outage probability of CORS and LBRS strategies tend to be consistent.The diversity order of the LBRS and CORS strategies both depends on the computing capabilities of the relay nodes,and the diversity order of the CPORS strategy is constant.Combining MEC and D2 D communication technology,using idle computing resources of terminal equipment to assist other task equipment to complete tasks offloading and computing,this paper studies the multi-user D2 D communication nodes selection strategies,and also extends the one-tomulti nodes selection strategies to the multi-to-multi nodes selection strategies.Traditional research focuses on the impact of communication resources such as spectrum multiplexing and interference management on D2 D communication networks on the nodes selection results.This paper aims to ensure the minimum bandwidth resource consumption at the tasks offloading rate,and studies nodes pairing strategies when terminal devices shared computing resources,energy resources,and communication link delays are all limited.Because the optimization problem is an NP-hard problem,this paper proposes a fully connected algorithm based on the Hungarian algorithm and the minimum distance greedy algorithm based on the greedy algorithm.The fully connected algorithm achieves lower time complexity by changing space to time.The minimum distance greedy algorithm preferentially connects the closest resource device and task device.It cannot guarantee that the system task can be completely offloading and it has a low space complexity.Simulations show that,under the low signal-to-noise ratio,the fully connected algorithm can still find the global optimal allocation with the task completely offloading but the completion rate of the minimum distance greedy algorithm will be affected by resource constraints.With the task offloading rate being 1,the fully connected algorithm still has a slight advantage,but the gain is limited.Therefore,when the system's task offloading rate has a higher priority,compared with the minimum distance greedy algorithm,the fully connected algorithm has an absolute global advantage.