Mobility and Congestion in Dynamical Multilayer Networks with Finite Storage Capacity

Multilayer networks describe well many real interconnected communicationand transportation systems, ranging from computer networks to multimodalmobility infrastructures. Here, we introduce a model in which the nodeshave a limited capacity of storing and processing the agents moving overa multilayer network, and their congestions trigger temporary faultswhich, in turn, dynamically affect the routing of agents seeking foruncongested paths. The study of the network performance under differentlayer velocities and node maximum capacities reveals the existence ofdelicate trade-offs between the number of served agents and their timeto travel to destination. We provide analytical estimates of the optimalbuffer size at which the travel time is minimum and of its dependence onthe velocity and number of links at the different layers. Phenomenareminiscent of the slower is faster effect and of the Braess' paradoxare observed in our dynamical multilayer setup.