Measuring in Round Trip

Also implemented in the Tangram-II Traffic Generator, the measurement in Round trip way allows the user to estimate parameters in this probe generation model. As shown in Figure 4, using this technique it is possible to estimate many measures of interest, but in this case they refer to the coming and going of the probes. Figure shows a PMF estimation of the round trip time from several probes sent from UFRJ and echoed by UMASS to the sender again.

Figure: Round Trip Delay from packets generated at the same instant of the One-way.

The initial purpose of implementing the round trip module was to compare its results with the one generated by the Two one-way set. However, several metrics may also be estimated using this measurement.

We suppose that generating both measurements together may give us an insight about the accuracy of the one-way estimation. In this way, we believe that measuring the round trip delay and the OWD at the same time makes it possible to avaliate the correctness of the Offset and Skew removal. To consider a correct estimation, not only the mean delay obtained from the round trip measurement and the sum of both one-way measurement in oposite directions should be close, but also the convolutions of both PMF curves estimated by one-way delay must be close to the curve generated by round trip delay. Figure shows in (A) the closeness of both PMF estimations (RTT and convolution function).

Figure: Comparison between Delay calculations: (A) PMF, (B) CDF.

Figure also shows in (B) and (C) the closeness of both curves, but in this case in the form of the cumulative distribution function (CDF). In (c) a log scale is applied to the $y$-axis.