Software-Defined Networking Control for X-haul Optical Networks in Testbed Experiments and Emulation

Abstract

Today’s telecommunication networks encounter challenges with rapidly growing traffic demands in various internet applications and services, such as video streaming, augmented/virtual reality, connected vehicles, dense wireless radio nodes, and edge cloud computing. Cloud radio access networks (C-RANs) have been proposed to enable resource sharing, modular radio functions, network scalability, and efficient energy management for future mobile wireless networks. In C-RANs, traditional co-located baseband units (BBUs) and radio units (RUs) are split into central units (CU) hosting BBU pools and massive numbers of RUs connected through fronthaul (FH) optical transport links. However, the communication between CUs and RUs using either digital transmission with the common public radio interface (CPRI) or analog transmission with radio-over-fiber requires high bandwidth and strict synchronization delay limits. Thus, the evolution of next-generation optical transport systems is required to build efficient, dynamic, and scalable communication networks that support data transmission with high capacity and ultra-low latency to realize high performing C-RAN architectures. Conventional commercial optical transport systems in metropolitan areas are based on wavelength-division multiplexing (WDM) networks where static wavelength channels are provisioned along fiber links between network nodes (containing optical switches or amplifiers) to ensure the data transmission of the peak traffic for backhaul (BH). This results in inefficient utilization of optical network resources in C-RANs where high-capacity and low-latency x-haul (FH, midhaul, and BH) optical transport is required. In addition, conventional optical network elements (NEs) with vendor-specific operating systems (OS) increases the cost of upgrading the system for higher performance, and the complexity of designing novel control planes for scalable networks. To address these problems, there is growing interest in optical transport networks built with open and fully-programmable optical systems using software defined networking (SDN) controlled white-boxes such as reconfigurable add/drop multiplexers (ROADMs), optical circuit switches (OCSs), and erbium dopped fiber amplifiers (EDFAs). This thesis examines SDN control strategies for x-haul optical systems in 5G and beyond wireless radio access networks. First, the Cloud Enhanced Open Software Defined Mobile Wireless Testbed for City-Scale Deployment (COSMOS) advanced wireless testbed is reviewed. A dedicated multi-functional Ryu SDN controller is implemented in the testbed’s optical network with wavelength channel assignment and topology reconfiguration for intra-/inter- domain control, network element (NE) monitoring, and a wireless handover experiment. Secondly, a BBU pool allocation optimization algorithm and a physical impairment-aware routing and wavelength assignment (PIA-RWA) considering midhaul BBU-RU functional split are explored to maximize traffic capacity and minimize resource occupation in an optical network of a New York metropolitan area C-RANs use case. In addition, several artificial neural network (ANN) models are also investigated to contribute accurate quality of transmission (QoT) prediction tools of the physical optical layer. Lastly, Mininet-Optical is developed as an extension to Mininet to achieve a novel multi-layer network emulation tool for SDN controller development. A dynamic optical SDN controller with least-congested PIA-RWA and BBU resource load balancing strategies is evaluated to enhance the network capacity in a virtual COSMOS environment emulated by Mininet-Optical considering various diurnal wireless traffic patterns.