Personalized QoE Improvement for Networking Video Service
Yun Gao,Xin Wei,Liang Zhou
DOI: https://doi.org/10.1109/jsac.2020.3000395
IF: 16.4
2020-10-01
IEEE Journal on Selected Areas in Communications
Abstract:Personalized networking video service, as an inevitable trend recently, has become the core part for users' quality of experience (QoE) improvement. Unfortunately, existing schemes of QoE improvement are far from personalized since most of them only focus on network-level or user-level optimization. How to realize personalized QoE improvement for networking video service has been widely considered as a fundamental technical challenge. To get over this dilemma, this work proposes a personalized QoE improvement scheme by fully taking advantage of the time-varying influences on users' QoE, including user-awareness, device-awareness and contextawareness. The highlights of this work lie in that, the proposed scheme realizes the personalization comprehensively considering all these three dimensions, meanwhile, it is so robust that can be applied to the application scenario where the observable users' data is not sufficient. Specifically, we firstly design a comprehensive data collection strategy and accurately classify these collected data. Then, an efficient deep learning (DL)-based model for personalized characteristics extraction is proposed to precisely characterize personalization with temporal, spatial and periodic correlations. Subsequently, to resolve the data sparsity issue, a federated learning (FL)-based architecture with privacy-protection is designed by securely exchanging encrypted parameters with other users. Importantly, we design an optimization scheme based on comprehensive MOS formula for personalized QoE improvement. Experimental results demonstrate that the proposed scheme has a significantly better performance on the personalized QoE improvement.
telecommunications,engineering, electrical & electronic