A Network Traffic Measurement Approach in Cloud-Edge SDN Networks

Abstract

Edge computing is a supplement to cloud computing. It is deployed at the edge of the access network and is closer to where data is generated and used. In 5G and future networks, a large number of devices dynamically access the network and integrate them into cloud computing for deep processing and have high requirements for transfer rates and response time. However, network performance is the bottleneck of the collaboration between cloud computing and edge computing. Network traffic measurement is the core of network traffic management. In order to solve the problems of low utilization of network resources and high difficulty in network management, we study the problem of network traffic measurement in cloud edge computing networks based on software-defined networking (SDN). We propose a new cloud edge network traffic measurement method based on SDN. In this method, we extract statistical records coarse-grained from OpenFlow switches and use them to train an autoregressive moving average (ARMA) model. Use the ARMA model to make fine-grained predictions of network traffic. In order to reduce the estimation error, we propose to use optimization methods to optimize the estimation results. However, we found that the objective function is a very difficult NP-difficult problem, so we use a heuristic algorithm to quickly find the optimal solution. Finally, we repeat some simulations to evaluate the proposed method.