Framework Establishment of Methodologies for Statistical Causal Inference in Multimedia Communications QoE

2022 Fiscal Year Final Research Report

• Project/Area Number: 20K04495
• Research Category: Grant-in-Aid for Scientific Research (C)
• Allocation Type: Multi-year Fund
• Section: 一般
• Review Section: Basic Section 21020:Communication and network engineering-related
• Research Institution: Nagoya Industrial Science Research Institute
• Principal Investigator: Tasaka Shuji 公益財団法人名古屋産業科学研究所, 研究部, 研究員 (80110261)
• Project Period (FY): 2020-04-01 – 2023-03-31
• Keywords: マルチメディア通信 / QoE / 統計的因果推論 / ベイズ統計モデル / 構造方程式モデル / MCMCシミュレーション / インターネット高度化

Outline of Final Research Achievements

This project aims at establishing a framework of Bayesian methodologies for statistical causal inference of QoE (Quality of Experience) in haptic audiovisual interactive communications.
We first proposed an empirical method for inferring causal directions of constructs in multidimensional QoE by structural equation models (SEMs) each with three constructs. Six SEMs with different directions were compared to show the one from the domain knowledge more plausible. We noticed that QoE can behave like the "impact-perceive-adapt" model reported by the University of Manchester in 2007.
We next studied how QoS (Quality of Service) affects prediction accuracy of multidimensional QoE by taking into consideration the causal direction of QoS to QoE. We proposed SEMs and MIMIC (Multiple Indicator MultIple Cause) models, which were compared with traditional logistic regression models. We found that multiple QoE measures should be utilized jointly rather than resorting to QoS information only.

Free Research Field

工学

Academic Significance and Societal Importance of the Research Achievements

因果関係モデルは，全ての学術分野における基盤である．しかしながら，インターネットにおけるマルチメディア通信QoEについては，その基礎的な研究さえもなされていなかった．送信側と受信側とで因果関係が異なりうることのメカニズムの解明や数理的モデルの構築は，因果推論の学術分野を拡大させる．更には，高度情報ネットワーク社会の安全かつ円滑な発展のために有用である．
また，力覚・音声・ビデオ通信は有望な次世代通信技術であり，そのQoE数理モデル構築はこの技術の発展に大いに役立つものと考える．