ARIMA-Based Traffic Forecasting for Quality of Service (QoS) Flow Routing in Sixth Generation (6G) Networks

We develop a methodology for Quality of Service (QoS) flow routing in IoT networks targeted at improving the performance of QoS flow optimization. In contrast with the existing schemes, the AutoRegressive Integrated Moving Average (ARIMA) model forecasts the Enhanced Mobil Broadband (eMBB) flow traffic periodically and a QoS optimization program makes reservations on the links for the predicted eMBB flows in advance in order to maximize the total end-to-end data rate from sources to destinations. To this end, we design an algorithm that delays the transmission of QoS flows that have laxer latency constraints. We compare the performance of our algorithm against a benchmark QoS routing scheme that transmits the flows in an on-demand fashion. Our results show that our algorithm outperforms the reactive approach. This work holds the potential to impact the design of QoS routing in beyond Fifth Generation (5G) networks.