Testing traffic congestion by distributed protocol analysis

In this paper, we propose the concept and the means for practical testing of TCP congestion window. With this respect, we present the architecture of the hardware and expert-system-based distributed protocol analysis that we used for practical data acquisition and testing, which we conducted on a major network with live traffic (Electronic Financial Transactions data transfer), as well as the appropriate algorithm for estimating the actual congestion window size from the measured data - mainly decoding with precise time-stamps (100 ns resolution locally and 1 μs with GPS clock distribution) and expert-system-processed results of the accordingly filtered data from the special-hardware-based capture buffer. We used statistical analysis for evaluation whether the data belonged to the specific (in this case assumed normal) cumulative distribution function, or whether two data sets exhibited the same statistical distribution - the condition-sine-qua-non for a stable interval. Finally, as the measured data-based congestion window values exhibited fitting, with satisfactory statistical significance, to the truncated normal distribution, we applied the appropriate model for estimation of the relevant parameters of the congestion window distribution: its mean value and variance.