Decentralized optimal traffic engineering in the Internet

Distributed optimal traffic engineering in the presence of multiple paths h as been found to be a difficult problem to solve. In this paper, we introdu ce a new approach in an attempt to tackle this problem. This approach has i ts basis in nonlinear control theory. More precisely, it relies on the conc ept of Sliding Modes. We develop a family of control laws, each of them hav ing the property that the steady-state network resource allocation yields t he maximum of the given utility function, subject to the network resource c onstraints. These control laws not only allow each ingress node to independ ently adjust its traffic sending rate but also provide a scheme for optimal traffic load redistribution among multiple paths. The only nonlocal inform ation needed is binary feedback from each congested node in the path. Moreo ver, the algorithms presented are applicable to a large class of utility fu nctions, namely, utility functions that can be expressed as the sum of conc ave functions of the sending rates. We show that the technique can be appli ed not only to rate adaptive traffic with multiple paths, but also to assur ed service traffic with multiple paths. Preliminary case studies show that this technique is potentially very useful for optimal traffic engineering i n a multiple-class-of-service and multiple-path enabled Internet, e.g., dif ferentiated services enabled multi-protocol label switching networks.