MOS Prediction for Mobile Broadband Networks Using Bayesian Artificial Intelligence

Abstract

Mobile broadband (MBB) networks are growing fast with supporting high-speed internet access. Fifth-generation networks promise an enhanced MBB that offers a high-speed data rate and video streaming with ultra-low latency. Thus, monitoring the level quality of these services supported by network providers becomes essential. Mobile network operators continuously optimize their network performance to provide a better quality of service and quality of experience. Moreover, artificial intelligence has been used considerably in optimizations to efficiently meet the requirements of future mobile networks. In this paper, we propose a Bayesian network model to predict the minimum opinion score (MOS), which contributes to evaluating the network performance of video streaming services. The proposed model depends on several input data, namely, bite rate, stalling load, and round-trip time. The predicted MOS depends on prior probability distributions to generate posterior probabilities. The predicted MOS depends on these input data. Results demonstrate that the proposed model achieves a high prediction accuracy of 86%, with a mean square error of 0.34. The proposed model also has a robust performance design through various testing methods.