Physically aware agile optical networks
With the development of new laser sources, fiber amplifiers, and other optical components, optical communication systems have undergone enormous growth and evolution in recent decades. The current trend of optical networking is to move towards dynamic, all-optical networks. In all-optical networks, information signals are transmitted from source to destination totally in the optical domain, without the usual optical-electrical-optical conversions at intermediate nodes. New challenges and opportunities emerge in different layers of the optical network architecture in this transition process. This research work explores several interesting topics in both the physical layer and the network layer of all-optical networks. Our studies examine physical impairments which can adversely influence network performance. Both novel proactive and reactive approaches are proposed to improve network performance and provide quality of service (QoS) for users. In the physical layer, network transients, including switch transients and amplifier transients, can pose a serious threat to signal quality in dynamic networks. In all-optical networks, these transients can escalate along a lightpath. In this research, a new functionality is added to the backward reservation protocol to eliminate switch transients and a power shaping technique implemented at the link layer is proposed to decrease in-line amplifier transients. Compared to other approaches in the literature, our designs are more general, economical, and can seamlessly cooperate with other solutions. In the network layer, a new QoS framework is proposed to provide QoS assurance in all-optical networks. The framework has two parts: the Physically Aware Routing algorithm (PAR) and the Physically Aware Backward Reservation protocol (PABR). Analytical models are incorporated into the QoS framework to predict lightpath signal quality with fiber nonlinear effects and ASE noise. For a connection request, the source node executes PAR to select a set of candidate paths which can possibly satisfy the user QoS requirement, and then starts PABR to probe candidate paths in parallel. The destination node selects a satisfactory lightpath from candidate paths. New functionality is designed in PABR to guarantee the signal quality of a lightpath during its life time. The proposed QoS framework is more efficient, scalable, and flexible compared to other benchmark algorithms.