We analyze aggregated traffic dynamics obtained from strongly linked network communities. Our results based on two empirical data traces from university campus networks indicate that neglecting the statistical links between traffic patterns generated by individual network nodes leads to the drastic underestimation of both waiting and sojourn times. We also show that similar effects can be observed in simulated traffic patterns obtained by agent based modeling. Moreover, we suggest several indices that could be used to quantify the links between nodes and show their relation with the queuing system performance indicators.
|Journal||Journal of physics: Conference series|
|Publication status||Published - 4 Nov 2019|
|Event||International Scientific and Practical Conference on Mathematical Modeling, Programming and Applied Mathematics 2019, MMPAM 2019 - Veliky Novgorod, Russian Federation|
Duration: 27 Jun 2019 → 28 Jun 2019