Autonomic Networking - from theory to practice

Autonomic Networking – from theory to practice

The first simulation of the first configuration is taken as the baseline, and the remaining simulations and all the other configurations are compared to this baseline response. The output signals of interest are chosen as engine torque, throttle, Autonomic Networking – from theory to practice and vehicle speed, to characterize the transparency from the different perspectives of the engine, driver, and vehicle, respectively. Before the statistical transparency analysis given in Section is applied, the raw data need to be aligned with respect to a common reference point. This point is taken as the instant the throttle command exceeds a threshold which is considered to mark the beginning of the simulation. Data collected before this point are ignored, which removes the data collected when the engine is idling before the simulation starts.

The analysis given in Section is then performed. Tables summarize the standard deviations in the selected output signals as defined by The one-way Autonomic Networking – from theory to practice ANOVA tables generated from these data are given in The extremely low p-values in all three ANOVA tables indicate that the null hypothesis that all configurations are the same for the signals considered is rejected with a significance level of. In other words, the variations of the signals show a statistically significant change in Autonomic Networking – from theory to practice at least one of the configurations. Thus, in at least one of the configurations there is a statistically significant loss in transparency Autonomic Networking – from theory to practice.

Autonomic Networking – from theory to practices

To obtain more specific results and to see how significantly each configuration affects the variation, pairwise comparisons for significant differences between the means of standard deviations of the signals are performed using Tukey’s method as proposed in Section. The results are visualized in, where the circles indicate the estimated mean standard deviations and the horizontal lines indicate the confidence intervals for the estimates. Overlapping confidence intervals are indicators of insignificant difference. The conclusion from is that the configuration pairs , are significantly different from each other, with a significance level for both engine torque and throttle data. This has a very important implication Autonomic Networking – from theory to practice that LAN- and Internet-distributed configurations are not significantly different from each other for these two signals. In otherwords, the change in transparency of engine torque and throttle is insignificant when the configuration moves from LAN distribution to Internet distribution with nominal delay.

Thus, the main source of degradation in transparency in this case is distributed simulation, and unless the Internet delay increases dramatically, it does not affect the transparency with respect to these two output signals significantly. When performed for the vehicle speed signal, Tukey’s test concludes that the significantly different pairs of configurations are , with a significance level of. Thus, another very important conclusion can be drawn from those results; namely, the change in variation, and thus, transparency, of vehicle speed is insignificant between Ideal, LAN-distributed, Autonomic Networking – from theory to practice and Internet-distributed configurations. Hence, as far as vehicle speed is concerned, the system can be considered transparen until the Internet delay increases dramatically. Autonomic Networking – from theory to practice.