Bandwidth estimation is the hot topic of the research. There are many bandwidth estimation tools but we argue that the evaluation environments of those tools, especially cross-traffic, are not the same. So the evaluations are not fair. In this paper, we first survey algorithms of various methods and cross-traffic features they used in their evaluation. Then we demonstrate the characters of cross-traffic are crucial to the estimation results. We propose a novel cross-traffic generator which can generate Internet-like traffic and can be easily implemented in the simulation environment. Preliminary test shows that our generator can be used in Internet relative algorithms evaluations.
Key words: Bandwidth Estimation; Cross-traffic
摘 要: 端到端的帶寬測量技術一直是研究的熱點。現在有很多測量帶寬的方法和工具，但是它們的測試環境特別是背景流量的設置，都是不相同的。首先調查了各種方法提出的測試環境；然后通過實驗證明了不同的背景流量會影響到測試結果；最后提出了生成公平背景流量的新方法。該方法具有實際網絡流量的特性，也便于在模擬環境中實現，可作為測量帶寬算法的評估標準。
關鍵詞: 帶寬檢測; 背景流量