Interruption probability of wireless video streaming with limited video lengths

    Research output: Contribution to journalArticleScientificpeer-review

    17 Citations (Scopus)


    In this paper, we consider a simple queueing theoretic method to predict the video interruption probability for a given video length. Specifically, a mobile user is streaming a video with a limited length and variable bit rate video encoding. The playback interruptions are caused by random packet delays occurring in the wireless link between the source and destination nodes. The dynamics of the playback buffer in the user terminal is modeled as a G/G/1 queue. To evaluate the video interruption probability, a simple asymptotic method has been presented for the case in which the video length approaches infinity. However, in many practical cases, the video length is limited, hindering the usage of the asymptotic method. We obtain a simple and closed-form upper bound for the analysis of the interruption probability that incorporates the effect of finite video lengths with known statistical delay parameters. Furthermore, a useful method is presented to select between the proposed method and the asymptotic method whose relative accuracy changes with the video length and statistical properties of the buffer load size. The accuracy of the proposed analytical method is compared with the existing methods. Finally, we address some practical challenges in buffer dimensioning when the statistical delay parameters are unknown and estimated with a finite number of received packets.
    Original languageEnglish
    Pages (from-to)1176-1180
    Number of pages5
    JournalIEEE Transactions on Multimedia
    Issue number4
    Publication statusPublished - 2014
    MoE publication typeA1 Journal article-refereed


    • finite video length
    • playback interruption
    • queueing theory
    • wireless video streaming


    Dive into the research topics of 'Interruption probability of wireless video streaming with limited video lengths'. Together they form a unique fingerprint.

    Cite this