Analysis of Control Injection Schemes for Free-Space Photonic Switching Architectures

All photonic switching architectures require control information to be injected into the nodes of the network fabric to permit appropriate routing of the traffic. However, there are several different schemes that can be used to perform this fundamental task. Schemes that have been proposed in the literature include the use of distributed control with self-routing packet headers, centralized control with self-routing packet headers, centralized control with spatial light modulators (SLMs), centralized control with direct node injection, and centralized control with individually-controlled surface emitting laser (SEL) arrays. In addition, hybrids combining several of these approaches can also be proposed. The manner in which control is injected into the network has a large impact on the performance and operating capabilities of the network, because the control scheme will affect the complexity of the switching node, the complexity of the optical hardware module, the system's tolerance to faulty nodes and to faulty power supplies (lasers), the required fabric bitrate, and the cost of memory for TMS functionality. In addition, for certain control schemes, the speed limitations, the pinout limitations, and the wire trace limitations on the memories, the SLMs, and the active switching devices may limit the overall size of the photonic switching networks. This paper will study these effects and compare the characteristics of several different control injection schemes.