Software Defined Networking-based Predictive Network Latency

In Software Defined Networking, the resources allocation is crucial for determining the performance, resource utilization and power consumption of the data plane, due to the significant bandwidth and latency requirements. In this article, we propose an efficient rules placement based latency measurement with online learning, that aims to dynamically predict the latency to update the flow rules in devices. The experimental results, using ONOS controller and Mininet, shows the efficiency of our algorithm in decreasing network latency.