IP網路群播影音串流品質參數之量測與分析

Abstract

本論文提出一個評估IP網路環境下，視訊串流品質的量測與分析方案。我們建立了MPEG-2 Transport Stream (TS) over UDP的群播影音串流模擬實驗環境，先由ETSI TR101290 MPEG-2 TS參數之端對端量測著手，於不同網路效能劣化條件下，探討MPEG-2 TS 品質效能指標 (Key Performance Indicator, KPI) 與人眼實際知覺感受之間的關連性。最後，我們利用ITU-T G.107 Annex B之模型進一步分析客戶滿意度與MPEG-2 TS KPI之間的相關性，據此得知TS KPI之品質劣化臨界值，以便及時反映客戶端視訊劣化現象。

在本研究中，我們建立視訊品質之量測模擬實驗環境，於影音串流伺服器與用戶端之間，利用網路劣化模擬儀器糢擬IP網路劣化環境，執行MPEG-2 TS量測程式量測各項主要TS參數，並由知覺視訊品質評分儀器獲得同時段該影音串流之MOS (Mean Opinion Score) 評分。再者，依據ETSI TR101290陳列之TS評估參數，以2sec量測取樣單位估算5min週期之劣化時間比(Error Ratio)，並研擬了數個TS KPI指標。於此實驗環境下實測網路劣化因子對視訊品質之影響、各個TS KPI與人眼實際知覺感受之相關性，並利用ITU-T G.107 AnnexB的方法，推演出MPEG-2 TS KPI與客戶滿意度(GoB,PoW)之關係，而得到TS KPIs之臨界值。

This thesis proposes a new measurement and analysis solution for evaluating video streaming quality over IP network. The author starts from measuring end-to-end quality based on ETSI (TR101290 MPEG-2 Transport Stream) parameters, under the impairment conditions of different network performance, and then discusses the relationship between MPEG-2 TS Key Performance Indicator (KPI) and human’s visualization. Finally, we utilize the model of ITU-T G.107 Annex B to further analyze the dependence between customers’ satisfaction and MPEG-2 TS KPIs , then find the appropriate thresholds of the KPIs to notify the worse effect of responding customer visualization in time. We set up a simulating experiment environment to measure multicast video streaming quality over UDP/IP network. Between video steaming server and customer-end, we used network error simulation instrument to simulate IP network error environment and execute the measurement program of MPEG-2 TS to measure all the major TS parameters. Through sense visual quality evaluation instrument, we obtained the Mean Opinion Score (MOS) of the simultaneous video streams. In this study, We set up a simulating experiment environment to measure and analyze the video quality. Between video steaming server and customer-end, we uses network simulation facility to simulate the disadvantage environment of IP network, executes MPEG-2 TS measurement programs to measure each major TS parameter, and obtains the MOS of video streams in the same period by using visualization quality evaluation facility. Furthermore, based on TS evaluation parameters displayed by ETSI TR101290, the author use 2sec to measure sample units in order to calculate the error ratio of 5min period, and then proposes several TS KPIs. Under this experiment environment, the author measures the influence of network-impaired factors on video quality and the correlation between each TS KPI and the sensibility of human eyes. The author also introduces the method of ITU-T G.107 AnnexB to the correlation between MPEG-2 TS KPI and customers’ satisfaction, then evaluate the threshold of the TS KPIs.