在點對點網路系統 (Peer-to-Peer systems, P2P) 中做資訊的流通散佈(Broadcast)是一件具有挑戰性的工作。由於P2P並不像傳統的主從式架構(client/server systems),由中央伺服器負責散發訊息,而是必須依賴互相連接的點之間密切溝通來達成,因此在過程當中經常出現大量的重複訊息在系統的點與點之間傳送,而造成了網路擁塞與額外負擔。這篇論文主要討論了在P2P系統的一般broadcast過程所隱含的理論基礎,進而利用所觀察到的數學模型來設計一個快速的broadcast架構。其中主要的概念是在系統尚未開始做溝通之前,根據模型給予每個點關於資訊傳遞過程的一些知識(preliminary knowledge),根據這個模型,每個點都能夠獨立決定如何將它收到的 訊息傳播出去,以幫助整個系統減少過多的重複。我們的實驗結果驗證了這樣的方式能夠使網路獲得最大效益,也就是使更多點快速的收到訊息,並且和所必須付出的代價之間取得平衡。
Broadcasting in peer-to-peer (P2P) systems is challenging because of the lack of central server and its dynamic characteristic. Peers usually waste too much unnecessary effort to forward duplicated messages. In this paper, we provide some theoretical basis for the broadcasting processes and utilize the theoretical model to design the broadcasting framework. The main idea of the framework is to provide some preliminary knowledge for each peer. Based on the preliminary knowledge, peers can give directions of broadcasting and eliminate the duplicated transmissions. Our experimental results show that in the dynamic systems of large scales, our method can achieve maximum performance that reaches the balance between the delivered time and traffic overhead.