Multi-step throughput prediction of 5G FR2 mobile channels

For all advantages of 5G mobile networks the new millimeter frequency band FR2 needs to be deployed. It has not yet been used for mobile services. The deployment of the FR2 band allows for the maximum speeds of 5G mobile channels but their throughput can wildly fluctuate over time. Some mobile applications, such as ultra high-definition video streaming, need to adapt for varying channel speeds on long time intervals. This adaptation can be realized with a multi-step throughput prediction of 5G mobile channels based on the previous measurements and some external factors. This paper explores the multi-step throughput prediction of 5G FR2 channels with an application of a wide range of machine learning methods.