Survivability Design in Elastic Optical Networks
Open Access DepositedThe rapid evolution of telecommunications, driven by the arrival of 5G technology, cloud computing, and the growth of data centers, has highlighted the need for high-capacity and flexible backbone optical networks. Elastic Optical Networks (EONs) have emerged as a critical innovation in this area, offering dynamic bandwidth allocation, improved spectral efficiency, and flexibility to support diverse and high-capacity traffic demands. As data usage continues to increase, particularly with the integration of 5G networks and the expansion of cloud services, EONs provide a promising solution to meet these challenges. As the dependence on digital services increases, ensuring the survivability of EONs becomes increasingly crucial. Survivability refers to the network's ability to maintain service continuity in the event of failures or disruptions, such as fiber cuts, hardware issues, or natural disasters. With the substantial demands placed on networks by 5G, cloud networks, and data centers, any downtime can result in significant data loss and service interruptions. Consequently, designing robust EONs with advanced fault management, quick recovery scheme, and effective resource allocation is essential. By enhancing the survivability of EONs, we can guarantee reliable and continuous service. To address these challenges, this dissertation explores various survivability designs in EONs. We first explore the p-cycle protection design in EONs. Due to the fast restoration and high protection efficiency, p-cycles have been extensively studied for conventional fixed-grid Wavelength-Division Multiplexing (WDM) networks. However, the p-cycle protection design for EONs was largely unexplored. We propose methods for evaluating and selecting p-cycles for both link protection (LP) and failure-independent path protection (FIPP) to survive single-link failures in translucent EONs with 3R regenerators. The proposed algorithms have better performance in terms of spectrum usage and blocking ratio. In addition to protection approaches, we also explore disaster recovery in EONs. Conventional protection techniques are typically designed for small-scale failures such as a small set of nodes and/or links. It is typically cost-prohibitive to directly deploy the existing protection mechanisms for disaster scenarios because of the huge amount of redundant resources that would be needed. To this end, we propose a series of disaster recovery algorithms with mitigation awareness (DRAMA). The concept of mitigation zone identifies a region surrounding the disaster zone wherein lightpaths may be reconfigured with degraded service in order to improve overall performance. We also consider machine learning-enabled disaster management, where the degraded service is selected by a deep reinforcement learning agent. The proposed designs outperform conventional recovery algorithms. Furthermore, we investigate disaster recovery in data center EONs (DC-EONs). With the network function virtualization paradigm, critical network functionalities can be instantiated as software-based virtual network functions (VNFs) deployed within the network's data centers. An end-to-end service can then be composed as an ordered set of VNFs called as service function chain (SFC). However, the potential data center failure makes the problem very challenging. We develop effective strategies for recovering SFCs after a disaster with the mitigation zone. SFCs within the mitigation zone are recovered with a relaxed latency threshold to address the congestion problem caused by a large-scale network failure. The proposed SFC recovery algorithm performs better than a baseline algorithm in several scenarios. Overall, this dissertation advances the survivability in elastic optical networks, by addressing both p-cycle protection and disaster management. These contributions are crucial for maintaining reliable service in elastic optical networks, meeting the increasing demands of applications in 5G, cloud computing, and data centers.
- All rights reserved
Notice to Authors
If you are the author of this work and you have any questions about the information on this page, please use the Contact form to get in touch with us.