The thesis presents methods to enable commercial off-the-shelf (COTS) datacom transceivers for use in optical circuit-switching applications. These applications require fast or (burst-mode) locking characteristics. This goal requires research in two areas. First, the burst-mode characteristics of (COTS) datacom transceivers were characterized as a function of the mean received power, the power offset between the channels being switched, and the transition time of the optical switch. Second, a high-speed board that can be used for variable optical attenuation was developed and tested to mitigate the power offset. This substantially improves the burst-mode lock times. The culmination of this work is a calibration procedure that can improve the locking characteristics of COTS transceivers used in a circuit-switched optical network.

Traditionally, datacenter networks are constructed with electronic packet switches, which forward data in a distributed manner. However, packet switches are running in to serious scaling challenges due to the massive growth in bandwidth requirements within major datacenters. One possible solution to this situation is to use optical circuit switching, which has significant potential for both power and cost savings.

However, circuit switching is a completely different paradigm from packet switching, and as such effectively utilizing circuit switching requires work at all layers in the network, including the network architecture, switches, transceivers, network interfaces, and protocols.

This dissertation validates the thesis that developing a NIC capable of precision admission control and characterizing its performance can lead to practical sub-microsecond circuit-switched networks at scale. Specifically, this dissertation presents novel contributions to two key system-level issues inherent in utilizing optical circuit switching in datacenter networking applications.

The first contribution of this dissertation quantifies the system-level reconfiguration time of an optically circuit-switched link. This work is critical for understanding how to build optical links using high speed optical switches and how to integrate these links into a datacenter environment. This includes a discussion of system-level reconfiguration time and link-level measurements of an optically switched link, including bit error rate (BER) characterization of a 25 Gbps link utilizing a burst-mode receiver, switched by a nanosecond-scale silicon photonic switch.

The second contribution of this dissertation is the development of a network interface controller (NIC) that can precisely control the injection of packets into an optically-switched network. This work describes and quantifies the performance of a novel high-performance, open-source, FPGA-based NIC called Corundum. This NIC is designed to precisely control the injection of packets from multiple queues into a circuit-switched network using a hardware-based scheduler. The platform provides the flexibility to implement high precision time synchronization as well as perform link-level characterization including BER measurements. The development of such a network interface can lead to practical sub-microsecond circuit-switched networks at scale.

Ultra-high capacity fiber optic systems with data rates exceeding 100 Giga bits per second per fiber are currently being deployed with higher capacity systems in development. The requirement of a minimum energy per bit for reliable communication means that the power launched into a single fiber is now at a level where significant nonlinearites exist. Nonlinearities can also be produced for lower power intensity-modulated systems because of the square-law nature of sensors.

In order to maximize the information capacity, the combined channel that includes a combination of nonlinear impairments along with additional linear impairments must be mitigated. This mitigation can be achieved by a combination of modulation coding at the transmitter and equalization at the receiver. The development of these techniques for nonlinear channels is significantly more complex than the corresponding techniques for linear channels because of the nature of the nonlinearity and the extremely high data rate. This rate limits the complexity of the equalization algorithm.

This thesis presents modulation coding and equalization techniques for several nonlinear fiber optic channels. We consider two classes of nonlinearity. The first arises from the combination of linear dispersion in an optical fiber and square-law sensing. The second arises from nonlinear propagation characteristics caused by a power-dependent index of refraction change called a Kerr nonlinearity.

A variety of nonlinear channel models can be constructed from these two fundamental forms of nonlinearity along with linear impairments. The dominant linear impairment is dispersion. One form occurs when the propagation characteristics for each mode depend on the frequency. A second form of dispersion arises because different polarization modes can have different propagation characteristics.

The research premise of this thesis is that a combination of modulation coding, sequence estimation (both single-user and multi-user) and nonlinear equalization based on heuristic algorithms can produce significant performance improvement relative to published techniques. We present several abstracted scenarios reflecting practical systems where a combination of these techniques is effective. We also describe situations where they are ineffective. These results lay the foundation for further work using these techniques to optimize specific nonlinear channels.