Generalized MPLS-based distributed control architecture for automatically switched transport networks

Abstract

Current circuit-switched transport networks, such as plesiochronous digital hierarchy (PDH) and synchronous optical network/synchronous digital hierarchy (SONET/SDH), have traditionally used centralized network management for connection control. To facilitate the value-added capabilities of today's networks — such as the rapid provisioning of services, dynamic setup of bandwidth requests, and fast mesh-based restoration — distributed connection control using signaling protocols has quickly gained industry momentum. Efforts have been initiated in various standards bodies to define the automatically switched transport network (ASTN). Although many architectural choices are now available, this paper describes a distributed control plane architecture that can be applied to various circuit-switching technologies and different network applications. This architecture adopts the concept of a generalized version of multiprotocol label switching (MPLS), which extends and modifies MPLS and other protocols on the Internet to make them applicable to various transport networks and also facilitates optical data networking. Four major functional components are incorporated in this architecture: element-level resource discovery, state information dissemination, path selection, and path control modules. Using these concepts, the transport network can be viewed as a virtual nonblocking, reconfigurable backplane of different network clients. This view represents a radical departure from the traditional data networking view of transport networks as providing fixed pipes and will have a dramatic impact on future network interworking and end-to-end traffic engineering (TE).