Simulation of backbone networks that rely on advanced optical switching techniques
Loading...
Date
2026-10-07
Authors
Ράπτη, Χριστίνα
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
The continuously growing needs of users over the years, combined with the rapid evolution of technology, have driven the enormous growth of optical networks and their fundamental contribution to supporting the global internet — despite the relatively recent emergence of such systems, roughly fifty years ago. Their importance, however, is becoming increasingly evident today, as demand for entertainment services continues to rise, and particularly intensely in recent years, with the corresponding surge in demand for cloud computing. To meet
these needs, while maintaining a satisfactory capital expenditure and at the same time achieving the desired performance, the primary goal appears to be preserving the existing infrastructure as much as possible, with the emphasis shifting towards optical planning and towards increasing capacity within already deployed fibers. Given the cost and complexity of actual deployment, such decisions must be evaluated computationally, through simulation, before any real infrastructure is committed. The aim of this thesis was to determine whether a Quality of Transmission (QoT)-aware optical network planning tool, the RMLSSA offline planning tool and the Q-tool, with Ultra-Wideband (UWB) transmission exploiting spectrum beyond the conventional C-band, can be run at realistic network scale. To that end, a 694-node network was first constructed from the data of the TEFNET24 reference network: the metro-regional segment of a broader backbone network, at the point where it interconnects with the national core. The computational bottleneck of the tool was then removed through demand-aware path precomputation, which computes candidate paths only for the node pairs that carry traffic, so that the tool runs quickly on conventional
hardware. Finally, two different traffic structures were evaluated on the same topology over seven years of doubling traffic: a synthetic traffic matrix constructed from real hourly measurements at three cities, and the traffic matrix of TEFNET24, consistent with the operator's own structure and scaled to the same volumes. In both cases no connection is blocked and resources never limit the network; transmission quality, however, does: with the synthetic matrix, a substantial share of the connections fails the target BER, growing over the years
and concentrated on long transparent lightpaths, whereas with the operator's matrix almost all connections meet it. Judged by blocking, throughput and modulation selection alone, the two scenarios would appear equivalent. Only the evaluation of transmission quality reveals the difference between them, as quality is governed by the structure of the demand rather than by its volume. Zero blocking is therefore not a sufficient planning metric, and QoT-constrained allocation and progressive band activation emerge as the natural continuation of
this work.
Description
Keywords
Multilayer hierarchical topology structure, Traffic matrix construction, Aggregation ring, Clustering analysis, Synthetic traffic matrix, Path precomputation, Quality of transmission, Optical signal to noise plus interference ratio, Average lightpath length, Blocking probability, Metro networks, Regional networks, Backbone networks