A Load-Balanced Agile All-Photonic Network

We investigate a load-balancing method in a time division multiplexed agile all-photonic network (AAPN), which has a star topology and a buffeHess optical core. This method for bandwidth sharing is derived from a packet switch architecture that consists of three electronic buffering stages holding layered cross-point queues and two optical transpose interconnects between the stages. For AAPN, the architecture is folded: the slots (the data units) are fust sent to the same set of edge nodes acting as intermediate stage, and then sent to their final-destination edge nodes. The approach is suitable for the metro / access scenarios in which propagation delays are small and it simplifies the scheduling problem significantly since it is fixed and performed locally, without the need for signaling or centralised schedulers.