UC Davis

As we continually churn out more and more information that must be transmitted over our networks, the networking challenges of handling this large-scale, highly-diverse, and varied- speed traffic opens exciting new research and development problems. More bandwidth and lower latency is required to accommodate the complex applications over wide-area networks. Although a complete traffic-adaptive network is still a long way to go, our networks are grad- ually incorporating changes through flexible network architectures such as Precision Time Synchronization Protocol (PTP), Elastic Optical Networks (EON), C+L bands expansion, Artificial Intelligence (AI), etc. This dissertation comprises of four contributions: i) high- precision time synchronization techniques for optical datacenter networks (Chapter 2); ii) dynamic resource allocation in mixed-grid optical networks (Chapter 3); iii) C+L bands up- grade strategies to sustain capacity crunch (Chapter 4); and iv) C to C+L bands upgrade with resource re-provisioning (Chapter 5). The dissertation concludes with a summary and future research directions (Chapter 6).The main topics of this dissertation are the following: 1. A datacenter, which is a highly-distributed multiprocessing system, needs to keep accurate track of time across a large number of machines. Precise time synchronization is critical due to stringent requirements of time-critical applications such as real-time big-data analytics, high-performance computing, Internet of things (IoT) networks, and financial trading. To achieve this time accuracy, we consider Precision Time Protocol (PTP) to synchronize the server clocks. Zero overhead is maintained by using data traffic to carry the time messages instead of a separate control channel. We showed that microsecond level of time accuracy can be achieved and discussed the dependency of the accuracy on different traffic loads, traffic distributions, and packet lengths. 2. A rapid change in traffic type, volume, and dynamicity is presented by cloud-based services, datacenters, 5G, Internet of Everything (IoE), etc. Although introduced as a promising solution to this change in 2008, EON technologies are not fully deployed yet. Rather a logical and gradual migration technique is adapted which investigates bottle- neck points of the network and takes migration decision while keeping the rest of the fixed-grid network operational. Therefore, a co-existing fixed-grid and flex-grid (which can be called a “mixed-grid”) is a cost-effective solution for current circumstances. However, this introduces new challenges by the interoperability issues between fixed and flex-grid technologies. We proposed a solution to a RSMA problem in a mixed- grid considering interoperability constraints. The solution proposes routes, spectrum, and modulation format to provision a dynamic, heterogeneous traffic on two US-wide network topologies ensuring maximum network throughput and minimum blocking. 3. As more traffic is put into the access network, the backbone network also needs to have higher capacity and better network planning. Although EON helped to maximize the available spectrum utilization in C band, the exploitation of bandwidth potential of single-mode fiber (SMF) can be achieved through opening up other spectrum bands. We investigate cost-efficient upgrade strategies for capacity enhancement in wide-area networks enabled by C+L bands. A multi-period strategy for upgrading network links from C band to C+L bands is proposed, ensuring physical-layer awareness, cost effec- tiveness, and less than 0.1% blocking. Results indicate that performance of an upgrade strategy depends on efficient selection of the sequence of links to be upgraded and on the time instant to upgrade, which are both topology- and traffic-dependent. 4. We study efficient allocation of resources during C to C+L bands network upgrade. After an upgrade, resource allocation may become sub-optimal, leading to lower utilization of spectrum resources causing requests blocking, early upgrade, and higher cost. Thus, we investigate pro-active re-provisioning of lightpaths after each upgrade for cost benefit. Our strategy locates highly-utilized links and upgrades them in batches. After each batch upgrade, existing traffic in C band is re-provisioned to L band. This re-provisioning frees up high-OSNR lightpaths in C band, leading to improved quality of future transmissions, delayed upgrades, and cost benefits. Results show that re- provisioning of a shorter lightpath provides the most cost-effective upgrade strategy.