A dual interior-point distributed algorithm for large-scale data networks optimization

This paper proposes a specialized decomposition algorithm, which incorporates the interior-point method into an augmented Lagrangian decomposition technique, for optimizing the quality of service through a data network. We show that many routing or resource allocation problems from the network literature can be recast into separable problems for which our algorithm can be applied. The proposed algorithm is highly parallelizable and thus suitable for distributively solving large-scale network problems.