Load Optimization Based on Edge Collaboration in Software Defined Ultra-Dense Networks

With the intelligence of user equipment and the popularization of emerging applications such as unmanned driving and face recognition, more and more computationally intensive and delay-sensitive tasks have been generated. As a new network paradigm, ultra-dense networks can greatly improve user access capabilities by deploying dense base stations (BSs). Edge computing can effectively guarantee the low-latency requirements of users in ultra-dense networks. However, the heterogeneity of servers, the distributed resources, and the dynamic energy consumption of mobile devices in ultra-dense networks make it extremely difficult for users to offload and load balance among servers. This paper applies the idea of software defined network to proposes an edge collaboration architecture to achieve resource sharing and efficient offloading of tasks based on the characteristics of global perception. In particular, considering the high load of the local server and the idle resources of the remote server, the best offloading strategy for users is obtained game theory. Simulation results show that the performance is improved by about 30% compared to the traditional local processing, edge offload and local edge random offload schemes.