Uncertainty-Aware optimization for Network Provisioning and Routing

In this paper, we propose a novel approach for the traffic engineering in computer networks with uncertain demand. By utilizing the different flexibility of network reconfiguration and characteristics of traffic demand in different timescales, we solve the network provisioning and routing problems through stochastic optimizations and their second-order cone programming approximations. Numerical illustration shows that our approach can reduce 25% of the network cost compared with the traditional traffic engineering solution.