Multi-Layer/Multi-Region Path Computation with Adaptation Capability Constraints

This paper addresses the Multi-Layer (ML)/Multi-Region (MR) optimal path computation problem in Generalized Multiprotocol Label Switching (GMPLS) networks. Within the context of traffic engineering, path computation consists of routing Label Switched Paths (LSPs) under a multitude of constraints while optimizing resource utilization. One such problem consists of finding a shortest path in a heterogeneous GMPLS network where the different link switching types and node switching adaptation capabilities present their own set of constraints. We argue that most work have overlooked these constraints perhaps due to an ambiguous interpretation of the switching adaptation functions defined by GMPLS. To this end, we propose a novel GMPLS path computation algorithm including a binary integer program (BIP) formulation which considers the complete set of ML/MR node adaptation constraints related to nesting, un-nesting and conversion.