Abstract

All photonic switching architectures require control information to be injected into the nodes of the network fabric to permit appropriate routing of the traffic. A typical photonic switch (Fig. 1-4) consists of nodes arranged in stages (node-stages) which are separated by optical links (link-stages). Every node in every node-stage must have control signals routed into it in addition to the data inputs and data outputs, and depending on the reconfiguration rate of the network, the aggregate bit-rate of the control inputs entering a node-stage can sometimes be as high as the aggregate bit-rate of the data passing through a node-stage. Thus, it should be apparent that the problem of control injection is not at all a trivial problem. Control injection is an important aspect of photonic switches that should probably by given more attention, because the manner in which control is injected into the network has a large effect on the performance and operating capabilities of the network. This paper will study these effects for four different control injection schemes: 1) distributed control scheme with self-routing packet headers, 2) centralized control scheme with self-routing packet headers, 3) centralized control scheme with spatial light modulators (SLMs), and 4) centralized control with direct node injection.

© 1991 Optical Society of America