結合以錯誤更正為基礎的遺失還原與多數群播團體來達到可擴展的可靠性群播傳輸

Abstract

在本論文中，我們研究一種結合以錯誤更正為基礎的遺失還原與多數群播團體的方法來達到可擴展的可靠性群播傳輸。以錯誤更正為基礎的遺失還原是一種搭配提前錯誤修正（FEC）與自動重傳要求（ARQ）的修復技術。在使用多數群播團體的技術中，其中一個群體用來傳輸原始的數據封包，而修正封包則傳送在不同的復原群體上。所有的接收者都是動態的加入或離開這些復原群體。我們顯示一個以錯誤更正為基礎的協定若能使用無限多個群播團體，比起那些僅使用一個群撥團體的協定，它能減少接收端的處理花費。接著，我們探究對於一個特定的協定而言，多少的群體才足夠讓接收端處理較少的無用修正封包。最後，我們描述兩種實現多數群體技術的方法。其中一種是局部過濾的設計，而另一種是根據需求控制的主動網路機制。

In this thesis, we investigate an approach that combines parity-based loss recovery with multiple multicast groups for scalable reliable multicast transmission. Parity-based loss recovery is a repair technique that integrates forward error correction (FEC) with automatic repeat request (ARQ). By the technique of multiple multicast groups, a single multicast group is used for the transmissions of original data packets, and the transmissions of repair packets are done to different recovery multicast groups, which all receivers dynamically join and leave. We show that parity-based protocols using an infinite number of multicast groups reduce the receiver processing overhead in comparison to the protocols using only a single group. Next, we explore how many groups are required to keep the processing of redundant packets at a receiver to a sufficiently low degree, for a specific protocol. Finally, we describe a local filtering scheme and a control-on-demand active networking mechanism for implementing the technique of multiple multicast groups.