Zhang Xiaoliang, Li Li
Information Engineering School Jiaozuo Normal College, Jiaozuo, 454000, China.
Heliyon. 2024 Apr 25;10(9):e30351. doi: 10.1016/j.heliyon.2024.e30351. eCollection 2024 May 15.
In the context of the burgeoning progression of wireless network technology and the corresponding escalation in the demand for mobile Internet-based multimedia transmission services, the task of preserving and augmenting user satisfaction has emerged as an imperative concern. This necessitates a sophisticated and accurate evaluation of multimedia service quality within the sphere of wireless networks. To systematically address the nuanced issue of user experience quality, the present study introduces a novel method for evaluating multimedia Quality of Experience (QoE) in wireless networks, employing an advanced deep learning model as the underlying analytical framework. Initially, the research undertakes the task of modeling the video session process, giving due consideration to the status of each temporal interval within the session's architecture. Subsequently, the challenge of QoE prediction is dissected and investigated through the lens of recurrent neural networks (RNNs), culminating in the proposition of an all-encompassing QoE prediction model that harmoniously integrates video information, Quality of Service (QoS) data, user behavior analytics, and facial expression analysis. The empirical segment of this research serves to validate the efficacy of the suggested video QoE evaluation method, engaging both quantitative and qualitative comparison metrics with contemporaneous state-of-the-art QoE models, employing the RTVCQoE dataset as the empirical foundation. The experimental findings illuminate that the QoE model elucidated in this study transcends competing models in performance metrics such as PLCC, SRCC, and KRCC. Consequently, this investigation stands as a seminal contribution to academic literature, furnishing an exacting and dependable QoE evaluation methodology. Such a contribution augments the user experience landscape in multimedia services within wireless networks, and instigates further scholarly exploration and technological innovation in the mobile Internet domain.
在无线网络技术蓬勃发展以及基于移动互联网的多媒体传输服务需求相应增加的背景下，保持并提高用户满意度的任务已成为当务之急。这就需要对无线网络领域内的多媒体服务质量进行复杂而准确的评估。为了系统地解决用户体验质量这一细微问题，本研究引入了一种用于评估无线网络中多媒体体验质量（QoE）的新方法，采用先进的深度学习模型作为基础分析框架。首先，该研究承担了对视频会话过程进行建模的任务，充分考虑会话架构内每个时间间隔的状态。随后，通过循环神经网络（RNN）对QoE预测的挑战进行剖析和研究，最终提出了一个全面的QoE预测模型，该模型将视频信息、服务质量（QoS）数据、用户行为分析和面部表情分析和谐地整合在一起。本研究的实证部分旨在验证所提出的视频QoE评估方法的有效性，将定量和定性比较指标与同期最先进的QoE模型进行比较，采用RTVCQoE数据集作为实证基础。实验结果表明，本研究中阐明的QoE模型在诸如PLCC、SRCC和KRCC等性能指标上超越了竞争模型。因此，这项研究是对学术文献的一项开创性贡献，提供了一种严格且可靠的QoE评估方法。这样的贡献改善了无线网络中多媒体服务的用户体验格局，并激发了移动互联网领域进一步的学术探索和技术创新。
