Making intelligent topology design choices: understanding structural and physical property performance implications in optical networks [Invited]

Abstract

The key goal in optical network design is to introduce intelligence in the network and deliver capacity when and where it is needed. It is critical to understand the dependencies between network topology properties and the achievable network throughput. Real topology data of optical networks is scarce and often large sets of synthetic graphs are used to evaluate their performance including proposed routing algorithms. These synthetic graphs are typically generated via the Erdos-Renyi (ER) and Barabasi-Albert (BA) models. Both models lead to distinct structural properties of the synthetic graphs, including the degree and diameter distributions. In this paper, we show that these two commonly used approaches are not adequate for the modelling of real optical networks. The structural properties of optical core networks are strongly influenced by the internodal distances. These, in turn, impact the signal-to-noise ratio, which is distance-dependent. The analysis of optical network performance must, therefore, include spatial awareness to better reflect the graph properties of optical core network topologies. In this work, a new variant of the BA model, taking into account the inter-nodal signal-to-noise ratio, is proposed. It is shown that this approach captures both the effects of graph structure and physical properties to generate better networks than traditional methods. The proposed model is compared to spatially agnostic approaches, in terms of the wavelength requirements and the total information throughput, and highlights how intelligent choices can significantly increase network throughputs whilst saving fibre.

Towards a Traffic-Optimal Large-Scale Optical Network Topology Design

“…To explore the relationship between the proposed DWC and the network throughput under different topology and demand scenarios, 10,000 topologies with 60-100 nodes were generated via SNR-BA generative graph model [12], 2,000 at each node scale. To systematically generate different demands, we skewed the demand distribution by randomly selecting half of the node pairs, setting the demand of them as (1−γ)λ while the other half as (1 + γ)λ, where λ is the demand value of a node pair under uniform demand distribution γ ∈ [0, 1] weighs how heavily the demand is skewed.…”

Ultra-fast Optical Network Throughput Prediction using Graph Neural Networks

“…A dataset of 80000 unique graphs, with 10 to 15 nodes, were used to train the MPNN. The node locations were chosen uniform randomly over a grid that represents the size of the north-American continent, from which the graphs were then generated via the SNR-BA model [5]. These were labelled with their corresponding maximum achievable throughput.…”

“…This is an evolution from the previous goals of minimising the number of wavelengths needed to optically route data within the network [3], where the relationship between wavelength requirements and the physical topology is well understood [4]. However, due to growing number of wavelengths in fibres and the associated linear and nonlinear physical layer impairments, physical properties play a significant role in determining both routing and throughput, and must be taken into account in network design [5].…”

Dynamic resource allocation in different ultrawideband optical network topologies

“…However, they were evaluated using a single topology. The network topology can have a significant impact on the overall performance and offered throughput, as shown in [5]. In this paper, we extend the study of the network topology's impact on an ultrawideband regime by evaluating the performance of recently proposed heuristics for different topologies.…”