A Model for Throughput Prediction for Mobile Users

In this paper we propose a stochastic model to predict user throughput in mobile networks. In particular, the model accounts for uncertainty such as random phenomena (e.g., fast fading) or inexact information (e.g., user location) to derive the statistical distribution of the user throughput. Such a model is highly useful for aiding scheduling and resource allocation decisions. In addition, we provide a taxonomy of prediction techniques to investigate error sources and the main characteristics of prediction accuracy. Finally, we show the versatility of the model by analyzing LTE user throughput for the case where knowledge of either the user's actual position or the congestion level in the cell is inexact.