Abstract
The development of Quality of Experience (QoE) models using Machine Learning (ML) is challenging, since it can be difficult to share datasets between research entities to protect the intellectual property of the ML model and the confidentiality of user studies in compliance with data protection regulations such as General Data Protection Regulation (GDPR). This makes distributed machine learning techniques that do not necessitate sharing of data or attribute names appealing. One suitable use case in the scope of QoE can be the task of mapping QoE indicators for the perception of quality such as Mean Opinion Scores (MOS), in a distributed manner. In this article, we present Distributed Ensemble Learning (DEL), and Vertical Federated Learning (vFL) to address this context. Both approaches can be applied to datasets that have different feature sets, i.e., split features. The DEL approach is ML model-agnostic and achieves up to 12% accuracy improvement of ensembling various generic and specific models. The vFL approach is based on neural networks and achieves on-par accuracy with a conventional Fully Centralized machine learning model, while exhibiting statistically significant performance that is superior to that of the Isolated local models with an average accuracy improvement of 26%. Moreover, energy-efficient vFL with reduced network footprint and training time is obtained by further tuning the model hyper-parameters.
Highlights
Quality of Experience (QoE) addresses the degree of user delight or annoyance [1].For service and network providers, it is important to control the factors that contribute to QoE, which are captured by QoE models
W0 = 0 indicates a scenario when model trained with G1 dataset is applied on the G0 dataset; while W0 = 1 represents the isolated scenario when only the model trained on G0 dataset is applied on the G0, no ensemble of models
We present two distributed machine learning techniques, Distributed Ensemble Learning (DEL) and Vertical Federated Learning (vFL), which can be used for collaborative model development on decentralized datasets with different feature sets. vFL in particular enables training machine learning models collaboratively between QoE research entities, which potentially benefits all collaborating entities mutually, even if the research entities have different feature sets
Summary
Quality of Experience (QoE) addresses the degree of user delight or annoyance [1]. For service and network providers, it is important to control the factors that contribute to QoE, which are captured by QoE models. End-to-end data collection, network performance monitoring and development of forecasting models for network performance prediction and/or QoE estimation, will be even harder due to the inherently distributed observations at different parts of the network path such as RAN (Radio Access Network), transport, and applications, which are expected to be provisioned by different business segments such as operators and service providers. All measurements performed on different segments need to be shared to compute overall end-to-end QoE This necessitates efficient and automated enablers such as distributed machine learning on multiple nonshared datasets that have potentially different feature sets. This way, a complete QoE model can be assembled leveraging end-to-end distributed single observations from different applications, networks, and energy performance sensors
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
