Optimal and Efficient End-to-End Path Computation in Multi-Layer Networks

This paper addresses the problem of end-to-end path computation in a transport network with multiple switching technologies, for which the label switched path (LSP) traffic engineering (TE) in the multi-layer networks is an important application. By transforming a network graph to a channel graph, we provide a novel and general solution to find optimal paths in multi-layer networks through vertically searching across layers and horizontally searching on the same layer. The channel graph yields an explicit view of the constraints associated with the nodes and links that otherwise are hidden in the network graph. The approach can be applied to constraints such as the wavelength continuity, encoding type, and switching bandwidth granularity. The proposed solution has been implemented in software and deployed in an experimental optical network.