Mapping of delay-sensitive virtual networks
Instead of at network interface level, such error checks are performed in the event-based framework by assigning unique event indices to all packets and using only the most recent ones. This ensures that both sites respond to the most up-to-date signals available. Mapping of delay-sensitive virtual networks Furthermore, a network characterization between Ann Arbor and Warren revealed that the packet drop rate is very low when using UDP. A typical result for round trip time delay versus timeof day for the network used in this paper is shown in its statistics are summarized in Table As Table I exemplifies, the drop rate was around for this network.
A packet is considered dropped in this case if it does not arrive within Such a low packet drop rate was preferred over increasing the delay by switching to TCP. also clearly shows a multimodal character in the sense that some packets experience a delay around and some around whereas others are dropped shown as zero delay in the figure. However, both the spikes and drops occur very infrequently as Table I shows, thereby further motivating the use of UDP. An increase in the network delay is simulated by bouncing the packets between Warren and Ann Arbor a number of timesbefore processing them. If the nominal round trip delay between Warren and Ann Arbor is approximately an effective network delay of for example, can be simulated by introducing additional round trips.
Mapping of delay-sensitive virtual network
Finally, it is worth emphasizing again that this communication architecture is chosen due to its ease of implementation and is not proposed or designed to give the lowest degradation in transparency possible or perform best in the sense of any other performance metric. This very communication architecture was also used in the human-in-the-loop experiments with the ID-HILS setup and it is desired in this work to stay as close as possible to the setup used in be able to generalize the conclusions to that setup, as well. Any transparency degradation due to this particular choice of communication architecture will be captured by the proposed transparency analysis. Mapping of delay-sensitive virtual networks Mapping of delay-sensitive virtual networks Four configurations are devised, each serving a different purpose. These are as follows. All models are integrated in a single simulation environment, simulated using a single numerical solver, in a computer that is directly connected to the engine. This configuration represents the baseline and is to help identify the variation inherent in the engine-in-theloop setup.
The models are distributed over the local area network LAN and co-simulated on two differen computers in Ann Arbor. This configuration is to help identify the variation introduced due to co-simulation using the adopted event-based framework. The delay introduced by theLANwas considered negligible. The models are distributed over the Internet, and the driver and vehicle models are simulated in Warren, while the driveline model is simulated in Ann Arbor and interacts with the engine. This configuration is to help identify the variation introduced due to distribution over the Internet using the event-based framework and with the nominal roundtrip time delay of approximately between Ann Arbor and Warren.The models are distributed over the Internet as in the configuration, however additional round trips are introduced into the communication protocol to increase the delay up to, a number representative of coast-to-coast delay in the U.S. This configuration is to help identify the variation introduced due to the additional delay. Mapping of delay-sensitive virtual networks Each configuration is simulated five times except for the first one, which is simulated six times.