Design and Service Provisioning Methods for Optical Networks in 5G and Beyond Scenarios

Doctoral thesis, 2023

Network operators are deploying 5G while also considering the evolution towards 6G. They consider different enablers and address various challenges. One trend in the 5G deployment is network densification, i.e., deploying many small cell sites close to the users, which need a well-designed transport network (TN). The choice of the TN technology and the location for processing the 5G protocol stack functions are critical to contain capital and operational expenditures. Furthermore, it is crucial to ensure the resiliency of the TN infrastructure in case of a failure in nodes and/or links while the resource efficiency is maximized.

Operators are also interested in 5G networks with flexibility and scalability features. In this context, one main question is where to deploy network functions so that the connectivity and compute resources are utilized efficiently while meeting strict service latency and availability requirements. Off-loading compute resources to large and central data centers (DCs) has some advantages, i.e., better utilization of compute resources at a lower cost. A backup path can be added to address service availability requirements when using compute off-loading strategies. This might impact the service blocking ratio and limit operators’ profit. The importance of this trade-off becomes more critical with the emergence of new 6G verticals.

This thesis proposes novel methods to address the issues outlined above. To address the challenge of cost-efficient TN deployment, the thesis introduces a framework to study the total cost of ownership (TCO), latency, and reliability performance of a set of TN architectures for high-layer and low-layer functional split options. The architectural options are fiber- or microwave-based. To address the strict availability requirement, the thesis proposes a resource-efficient protection strategy against single node/link failure of the midhaul segment. The method selects primary and backup DCs for each aggregation node (i.e., nodes to which cell sites are connected) while maximizing the sharing of backup resources. Finally, to address the challenge of resource efficiency while provisioning services, the thesis proposes a backup-enhanced compute off-loading strategy (i.e., resource-efficient provisioning (REP)). REP selects a DC, a connectivity path, and (optionally) a backup path for each service request with the aim of minimizing resource usage while the service latency and availability requirements are met.

Our results of the techno-economic assessment of the TN options reveal that, in some cases, microwave can be a good substitute for fiber technology. Several factors, including the geo-type, functional split option, and the cost of fiber trenching and microwave equipment, influence the effectiveness of the microwave. The considered architectures show similar latency and reliability performance and meet the 5G service requirements. The thesis also shows that a protection strategy based on shared connectivity and compute resources can lead to significant cost savings compared to benchmarks based on dedicated backup resources. Finally, the thesis shows that the proposed backup-enhanced compute off-loading strategy offers advantages in service blocking ratio and profit gain compared to a conventional off-loading approach that does not add a backup path. Benefits are even more evident considering next-generation services, e.g., expected on the market in 3 to 5 years, as the demand for services with stringent latency and availability will increase.