在行動同儕網路(Mobile peer-to-peer)環境中提供視訊串流服務面臨著許多傳輸上的挑戰,例如:高封包遺失率、網路連線不穩定與使用者網路的異質性等問題。因此本論文針對這些議題,提出了兩種創新傳輸策略:「統計式下載(Statistical Download;SD) 」與「無區塊式交錯(Blockless Interleaving;BI)」,藉以配合UEP-FEC通道編碼,提供調適性的視訊品質。SD是基於一個RTT機率模型,利用多來源的下載方式,提升封包下載成功率;而BI能根據來源端的網路遺失情形,提供不同的interleaving傳輸策略。模擬實驗結果發現,即使系統在高度異動與異質性的網路下,SD仍然可以提供穩定且即時的資料速率,同時BI亦能有效提升UEP-FEC解碼效能,降低封包叢集遺失的影響。而這些措施能明顯提升串流影片播放的連續性,減少中斷發生機率。
Provisioning of video streaming service in Mobile P2P networks is challenged by many facts in transmission, such as high data loss rate, burst loss, unstable connectivity, link and peer host heterogeneity. This thesis addresses these transport issues by proposing two novel schemes, statistical download (SD) and blockless interleaving (BI), which cooperatively work with UEP-FEC channel coding. SD improves the successful rate of data download by exploiting multi-source download according to a RTT-based probability model. BI conducts different interleaving policies of data download from respective source peers according to the associated channel loss process. Simulation results show that even the system presents high volatility and heterogeneity, when the size grows into large, SD can improve on-time data rate with moderate traffic overheads. Meanwhile, BI can improve the error recovery performance of UEP-FEC by eliminating the packet loss correlation. It turns out that the video continuity is significantly improved.