具服務品質之網狀網路傳輸流量優化機制之設計與實作

網路技術目前已發展出許多對於提高訊息傳送效率的方法，針對異質網路的設計也常見於現實部網中。其中洪水演算法為高可靠度的傳送方式，常見於廣播等使用。但在點對點(Peer-to-Peer, P2P)的網路中，此演算法會產生重複多餘的訊息而造成資源的浪費。本論文提出改良式的洪水演算法，有效改善了上述的問題。在不影響傳輸可靠度的情況下，提升了有效吞吐量，大幅降低傳輸成本。實驗的結果分析也有助於未來在已知設備的頻寬與數量下，可依其比例關係設計出具最佳有效吞吐量的異質網路架構，可更有效地利用網路資源。

關鍵字

洪水演算法；網狀網路；有效吞吐量

並列摘要

Current network technology has been able to offer transmission methods of higher efficiency. Heterogeneous network designs are also common in real-world networks. Among those designs, flooding algorithm is a highly reliable method of transmission, commonly seen in broadcasting. In peer-topeer (P2P) networks, however, this algorithm generates redundant repetitions of information, causing waste of resources. In this thesis a revised flooding algorithm is proposed, which effectively reduces redundant traffic for reliable message transmission in mesh networks. Without compromising transmission reliability, goodput is enhanced, significantly lowering cost of transmission. Analysis of the experiment results is also useful in designing a heterogeneous network structure with highest goodput based on the proportionality where equipment bandwidth and quantity are known, benefitting network efficiency.