I am running a simulation in Unetstack (https://unetstack.net/) with one protocol agent and one Poisson source agent (using PoissonBehavior) with a specified mean arrival delay, which sends DatagramReq() to the protocol agent according to the distribution.
I am comparing two different variants of Dflood (R. Otnes and S. Haavik, "Duplicate reduction with adaptive backoff for a flooding-based underwater network protocol," 2013 MTS/IEEE OCEANS - Bergen), and I noticed that for the variant that is exactly as described in the original reference the choice of min/max delays in the flooding rules affect the rate of the Poisson behavior. When running a simulation of a certain time length, I expect the source agent to generate a certain number of packets based on the average delay parameter. However, the parametrization of the protocol agent seems to affect the PoissionBehavior of the source agent, without them having any apparent connection other than the simulation framework that they are both running in, where the discrepancy is too large to be caused by a a random variation.
When running the other protocol, which functions differently, the source agent behaves as expected. Furthermore, if I replace the PoissonBehavior with a TickerBehavior in the source agent, the number of generated packets is as expected for both variants of the protocol agent.
I can also add that I am running DiscreteEventSimulator and that this happens with both ProtocolChannelModel and BasicAcousticChannel.
If the developers of Unetstack have any suggestions on how the PoissonBehavior could be affected by the behavior by another agent / other things happening in the simulation I would very much appreciate it.