Prediction of QoS Fulfillment for Notification Control in 5G Networks

In 5G, a new Quality of Service (QoS) parameter, Notification control, has been added for guaranteed bit rate flows. If enabled, it is expected that the Radio Access Network (RAN) will notify the Core Network if it is unable to fulfill its QoS requirements and that, if an interface exists, the notification could be forwarded to the application to allow it to adapt. In this study, methods for predicting Quality of Service (QoS) fulfillment for Notification control in 5G networks were investigated. Two Notification control algorithms were developed, one based on thresholds on queuing delays, and one based on if the users were expected to be allocated sufficient resources. To evaluate the algorithms, a definition for QoS fulfillment was proposed.The evaluation was performed on simulated data, where users transmitting fixed rate uplink video were investigated. The network consisted of nine cells in a hexagonal grid pattern, where each cell had a single 2600 MHz carrier. The network was stressed both by increasing the number of users, and by increasing the size of the cells. Logs of relevant metrics were extracted from the simulator and imported into MATLAB, where they were sampled at a fixed rate of 10 ms. The results showed that, on the simulated data, both algorithms could achieve missed notification rates below 10% while keeping the false notification rate below 50%. Furthermore, the time from a notification to a QoS non-fulfillment period was on the order of 100 ms. In conclusion, the tested algorithms could correctly trigger notifications inmost cases, but the application would have to adapt quickly. Further research should focus on proactive use-cases for Notification control, for example, by pairing Notification control with a congestion control algorithm. In such cases, Notification control could be used to give advanced warning to a user before giving it lower priority