Predictive dynamic multi-flow routing (PD-MFR) algorithm towards sixth generation (6G) software-defined networks

Abstract

We develop a dynamic Quality of Service (QoS) routing algorithm based on network traffic prediction for Sixth Generation (6G) SDNs. First we formulate a mixed integer optimization model that incorporates the key constraints for Ultra-Reliable Low Latency Communication (URLLC) enhanced Mobile Broadband (eMBB) and massive Machine-Type Communication (mMTC) traffic. Second we develop our Predictive Dynamic Multi-Flow Routing (PD-MFR) algorithm for QoS flows based on this optimization model. In PD-MFR first the network forms predictions of the aggregate eMBB traffic flow generation rates and makes reservations for the flows on the upcoming routing window. Second delay-tolerant mMTC flows are scheduled to be routed to fill up the residual capacities that remain after the eMBB flow reservations. Third URLLC flows are routed reactively. We demonstrate the performance of our PD-MFR algorithm when Autoregressive Integrated Moving Average (ARIMA) and Multi-Layer Perceptron (MLP) models are used in forecasting the eMBB flow generation rates. We measure the performance of PD-MFR against the benchmark QoS-Shortest Path Algorithm (QoS-SPA) in which all of the QoS flows are routed reactively and show that PD-MFR outperforms QoS-SPA significantly. This work advances the state of the art in QoS routing algorithms based on network traffic prediction geared towards next-generation SDNs.