在無線區域網路中運用環境仿真的真實流量重播

Abstract

真實流量重播的技術在有線網路上已經是一項成熟的技術，它擁有真實流量的高複雜度及重播可重複測試的特性，故廣泛用於測試網路產品的效能上。它主要的概念是透過精準控制流量撥出的先後順序來保持和DUT (Device Under Test)之間的互動。但要將流量重播的技術應用在無線網路上遭遇到很大的挑戰；因為無線網路上的通訊媒介既開放又缺乏保護，所有位於同一或鄰近頻道的裝置同時分享媒介的結果常造成干擾或雜訊影響接收訊號的正確性，且訊號在開放式空間中傳遞也面臨能量衰減的問題．故只有重播流量並不足以正確重製流量真實環境中的行為，也就無法達到用相同的情境測試DUT的目的。因此本論文提出一套以事件驅動重播的方法EDASR，在重播流量的同時也必須控制環境的影響來滿足重製的有效性．錄製流量時也同時收集環境的影響因素(訊號強度，雜訊強度，干擾)，然後在testbed上在環境重製的情形下重播流量.結果顯示reproduction ratio分別在DUT相依的流量和不理想的無線環境下仍能保持95.9%和88.7%的reproduction ratio。反之當沒有應用EDASR的情況下, reproduction ratio最差只有20.6%和0%的reproduction ratio。

Real traffic replay is one of the solutions to test network devices in labs. Using packet traces captured in the real environment could hold more details than modeling. However, packet replay control is required to manage the interactions with the DUT, and environment effects highly affect the packet exchange sequences, especially in wireless networks. Therefore, to apply traffic replay in wireless networks requires not only packet replay control but also the reproduction of environment effects. In this paper, we propose a method called event-driven automata-synchronized replay (EDASR) to address traffic replay. EDASR translates traffic traces into events following the IEEE802.11 protocol. The automaton of each event is applied to achieve the packet replay control with the environment effects. To evaluate traffic replay solutions, a performance metric, named reproduction ratio, is given to quantify how traffic replay approximates the traffic to the real environment. The software implementation of EDASR demonstrates that 95.9% and 88.7% reproduction ratios could be retained under DUT-dependent traffic and un-ideal wireless environment, respectively. Without EDASR, the worst cases of reproduction ratios are 20.6% and 0% respectively.