RWA decompositions for optimal throughput in reconfigurable optical networks

We consider an optical networking architecture with each node consisting of an electronic router (to model Internet Protocol [IP] functionality) overlaying optical crossconnect (OXC), with optical add/drop multiplexers (ADM) interfacing the optical and electronic layers, as depicted in Figure 1. Nodes are interconnected with optical fiber links, and the graph representation of these interconnections is termed the physical topology. OXC’s and ADM’s allow each node to source, terminate, or bypass a wavelength signal. Thus, a path across multiple fibers can be established for information flow between two nodes entirely at the optical layer, with conversion to and from the electronic layer only occuring at the source and termination of the path. Such a path is termed a lightpath. The graph consisting of the logical interconnections (source-destination pairs) implied by all active lightpaths is termed the logical topology of the network. In order to deal effectively with time-varying traffic, we equip our network with configurable components. In particular, we model each node with tunable transceivers and configurable optical switches. These tunable components introduce the flexibility to dynamically configure the logical topology of the network in response to variations in electronic layer queue occupancies. As a result, we can improve the throughput properties of the network by using algorithms employing joint optical layer reconfiguration and electronic layer routing.