Transition to congestion in communication/computation networks via Montecarlo simulations: from optimal resource allocation to real-world routing algorithms.
Marco Cogoni, Giovanni Busonera, Paolo Anedda, Gianluigi Zanetti · arXiv (Cornell University) · 2014
We generalize previous studies on critical phenomena in communication networks by adding computational capabilities to the nodes. A set of tasks with random origin, destination and computational structure is distributed on a network modeled as a graph and the latency of each task is computed during impulsive load simulations. The sum of all latencies is used as the energy in a Montecarlo simulation in which a near-zero temperature leads to optimal resource allocation whereas higher values mimic actual balancing algorithms. We study the transition to congestion by varying two parameters: system load (number of tasks) and temperature (resource assignment optimality). Finally, the time-evolution of the latency is approximately recovered by interpolating the latency probability distributions from the set of impulsive load simulations. The time-evolution of the system allows us to study the standard transition to the congested phase by varying the $\lambda$ parameter governing the Poisson task production rate. This approach allows us to reproduce, at least qualitatively, the main known results on network congestion and to gain a deeper insight over the maximum theoretical performance of a system.