Network decomposition and multi‐path routing optimal control

Abstract

In this work we face the Routing problem defined as an optimal control problem, with control variables representing the percentages of each flow routed along the available paths (multi‐path routing), and a set of cost functions corresponding to the two different objectives of traffic balancing and bottleneck load minimization. A key issue considered in this paper is the possibility of splitting the given network in sub‐networks, each one controlled by a separate subset of variables. This of course designs a decomposition of the original control problem in a set of easier‐to‐be‐dealt‐with lower dimensional problems. Some simulation results validate the procedure. Copyright © 2012 John Wiley & Sons, Ltd.

Distributed Control in Virtualized Networks

“…To offer to the overlying applications more complex network functionalities for the data management & analytics and for the resource configuration & control (e.g. routing 14 , load balancing 15 , admission control 16 , bandwidth on demand 17,18 , resource allocation 19,20 , quality of experience 21 , security 22 , etc. ), a NFV approach is proposed.…”

Coordinated throughput optimization for mobile sensor networks under heterogeneous fading conditions

“…The Fiedler eigenvalue is a good measure of graph connectivity in undirected networks; however, it should be noted that it does not truly reflect the quality of communication channel, especially in wireless networks. Another example of achieving network decomposition and multipath optimal routing is provided by Bruni et al 13 The authors has defined the routing problem as optimal control problem with control variables used to represent the flows routed on the available paths. The optimum positioning of routers is also helpful in minimizing the energy consumption.…”

Securing Cyber-Physical Systems: An Optimization Framework based on OSSTMM and Genetic Algorithms

“…In the proposed framework, the security maximisation problem will be modeled as a large-scale binary optimisation problem, making the choice of standard Genetic Algorithms natural, as they were proven to be both efficient and effective in such scenarios [38]. It is worth noting that more refined solution that benefit from other optimisation domains, as fuzzy logic control systems [39]- [41] or neural networks [42] and network/scenario decompositions [43], may prove useful for particularly large scenarios or when dealing with secu-rity metrics characterised by a more complex mathematical structure than OSSTMM.…”