As the interconnecting segment between access and the core networks, optical metro networks must accommodate ever-increasing and heterogeneous traffic demands, from low-rate demands (e.g., 1 Gbps to 10 Gbps) fueled by recent deployments of Fiber-to-the-Home (FTTH), to large demands supporting interconnection of the data centers (that can require up to 400 Gbps). Moreover, new services and applications are being introduced following the emergence of new technologies, such as smart cities, artificial intelligence, etc, that also contribute to the increase in the number of customers and requested capacity. Therefore, novel solutions to increase the capacity of optical metro networks are constantly sought after by vendors and operators worldwide. Such solutions must support recent coherent transmission technologies to support high transmission rates, but also ensure compatibility with the current metro transmission technologies, as e.g., legacy 10G non-coherent transmission technologies. Accordingly, new and scalable optimization approaches must be devised to achieve low-cost deployment of equipment and to efficiently allocate resources in the optical metro network. Different technologies can be considered as low-cost enablers at various layers of the optical metro networks. In this work, we focus on the combined application of three such enablers: traffic grooming, filterless node architectures, and optimized optical-amplifier placement. Traffic grooming consists in enabling the aggregation of low-rate clients (e.g., 1G and 10G) to higher-rate ones (e.g., 10G and 100/200/400G) using electronic grooming equipment at the Optical Transport Network (OTN) layer. Traffic grooming allows for reducing the number of OTN lines at the cost of increasing the number of grooming boards. The OTN layer can support different grooming boards that can enable traffic grooming in different patterns. As an example, the hierarchy of OTN boards can enable hierarchical traffic grooming, allowing various choices of granular traffic grooming or muxponders that can enable less granular traffic grooming but allow to reach higher rates and do not need separate grooming matrices as traditional boards, hence enforcing new types of constraints in the grooming problem. At the Wavelength Division Multiplexing (WDM) layer, filterless node architecture has been demonstrated to provide significant cost savings. In Filterless Optical Networks (FONs), the costly Wavelength Selective Switches (WSS) used in Wavelength-Switched Optical Networks (WSON) are substituted with inexpensive and unadorned passive equipment such as splitters and combiners. This replacement results in lower deployment costs and less power consumption. However, it causes spectrum waste since the Amplified Spontaneous Emission (ASE) noise generated by optical amplifiers circulates in the network even after reaching the destination due to a lack of WSS filtering, making the reuse of spectrum impossible. In FONs, the interconnection of the nodes is fixed and does not permit a flexible interconnection in case of dynamic traffic or traffic change, resulting in more constrained routing than WSON, which provides all-to-all node interconnections by leveraging WSSes. The significant limitation to the routing flexibility can be addressed using the so-called Programmable FONs (PFONs) that include in the node architecture some spatial switching element to enable at least fiber switching capabilities. PFONs provide an alternative to FONs, as they can still enable the low-cost benefits of the FON while partially limiting its drawbacks, as spatial switches can at least provide reconfigurable fiber-level interconnection of the nodes, hence achieving more flexible routing and OTN optimization. Moreover, at the WDM layer, an optimized placement of Optical Amplifiers (OAs) in metro networks can also offer opportunities for cost minimization. OAs are used to increase signal reach in WDM systems, enabling the simultaneous boost of multiple wavelengths and reducing the need for costly regeneration equipment. The optimized placement of OAs can guarantee low-cost optical metro architecture by removing unnecessary OAs, reducing power consumption, and maximizing signal quality. Proper placement of OAs allows to compensate signal loss while counteracting noise and nonlinear effects. Considering both traffic grooming, FON-related constraints and OA placement results in a complex and non-trivial joint logical and physical layer design. For example, OA placement can heavily influence the cost of the OTN layer in terms of grooming boards due to the interdependence of the OA placement and established lightpaths (and their bit rate) in the grooming solution. This thesis devises novel optimization approaches for both single-layer OTN to multi-layer OTN-over-WDM equipment cost minimization, considering different scenarios of traffic grooming (e.g., obtained using grooming boards vs. muxponders), filterless node architecture (e.g., FON vs PFON), and optimized OA placement, and considering backward compatibility with 10Gbps technologies. The proposed approaches are tested on real network topologies and traffic matrices provided by our industrial collaborators. This thesis explores single-layer and multi-layer optimizations by leveraging different heuristics, meta-heuristics, and Integer Linear Programming (ILP) methods. More precisely, the thesis presents four main contributions, two focusing on OTN-layer optimization and two focusing on cross-layer OTN-over-WDM optimization. First, we propose different optimization approaches to minimize the cost of OTN equipment at the OTN layer while deciding on the best-fitted grooming and routing solution. Second, we introduce a new equipment, called a muxponder, into the OTN layer to explore its impact on cost minimization. Third, we propose different multi-layer techniques for joint equipment cost optimization at the OTN and WDM layers. Finally, we investigate the impact of different node structures, such as FON, PFON, and WSON, on the multi-layer cost optimization.
In quanto segmento di interconnessione tra reti di accesso e core, le reti ottiche metropolitane devono soddisfare richieste di traffico sempre crescenti ed eterogenee, da richieste a bassa velocità (ad esempio, da 1 Gbps a 10 Gbps) alimentate dalle recenti implementazioni di Fiber-to-the-Home (FTTH), a grandi richieste che supportano l'interconnessione dei data center (che possono richiedere fino a 400 Gbps). Inoltre, nuovi servizi e applicazioni vengono introdotti in seguito all'emergere di nuove tecnologie, come smart cities, intelligenza artificiale, ecc., che contribuiscono anche all'aumento del numero di client e della capacità richiesta. Pertanto, nuove soluzioni per aumentare la capacità delle reti ottiche metropolitane sono costantemente ricercate da fornitori e operatori in tutto il mondo. Tali soluzioni devono supportare le recenti tecnologie di trasmissione coerente per supportare elevate velocità di trasmissione, ma anche garantire la compatibilità con le attuali tecnologie di trasmissione metropolitana, come ad esempio le tecnologie di trasmissione non coerente 10G legacy. Di conseguenza, è necessario ideare nuovi approcci di ottimizzazione scalabili per ottenere un'implementazione a basso costo delle apparecchiature e per allocare in modo efficiente le risorse nella rete metropolitana ottica. Diverse tecnologie possono essere considerate come abilitatori a basso costo a vari livelli delle reti ottiche metropolitane. In questo lavoro ci concentriamo sull'applicazione combinata di tre di tali abilitatori: traffic grooming, architetture di nodi senza filtro e posizionamento ottimizzato dell'amplificatore ottico. Il traffic grooming consiste nell'abilitare l'aggregazione di client a bassa velocità (ad esempio, 1G e 10G) a quelli a velocità più elevata (ad esempio, 10G e 100/200/400G) utilizzando apparecchiature di grooming elettroniche a livello di Optical Transport Network (OTN). Il traffic grooming consente di ridurre il numero di linee OTN richieste a costo di aumentare il numero di schede di grooming. Il livello OTN può supportare diverse schede di grooming che possono abilitare il traffic grooming in diversi schemi. Ad esempio, la gerarchia delle schede OTN può abilitare la gestione gerarchica del traffico, consentendo varie scelte di granulare traffic grooming o muxponder che possono abilitare una gestione meno granulare del traffico ma consentono di raggiungere velocità più elevate e non necessitano di matrici di gestione separate come le schede tradizionali, imponendo quindi nuovi tipi di vincoli nel problema della gestione. A livello WDM, è stato dimostrato che l'architettura dei nodi Filterless fornisce notevoli risparmi sui costi. Nelle reti ottiche senza filtro (FON), i costosi Wavelength Selective Switches (WSS) utilizzati nelle reti ottiche a commutazione di lunghezza d'onda (WSON) vengono sostituiti con apparecchiature passive poco costose e semplici come splitter e combiner. Questa sostituzione comporta minori costi di distribuzione e un minor consumo di energia. Tuttavia, causa spreco di spettro poiché il rumore di emissione spontanea amplificata (ASE) generato dagli amplificatori ottici circola nella rete anche dopo aver raggiunto la destinazione a causa della mancanza di filtraggio WSS, rendendo impossibile il riutilizzo dello spettro. Nelle FON, l'interconnessione dei nodi è fissa e non consente un'interconnessione flessibile in caso di traffico dinamico o di modifica del traffico, con conseguente routing più vincolato rispetto a WSON, che fornisce interconnessioni di nodi all-to-all sfruttando i WSS. La significativa limitazione alla flessibilità di routing può essere affrontata utilizzando i cosiddetti FON programmabili (PFON) che includono nell'architettura del nodo alcuni elementi di commutazione spaziale per abilitare almeno le capacità di commutazione in fibra. I PFON forniscono un'alternativa alle FON, in quanto possono comunque abilitare i vantaggi a basso costo delle FON limitandone parzialmente gli svantaggi, poiché gli switch spaziali possono almeno fornire un'interconnessione a livello di fibra riconfigurabile dei nodi, ottenendo quindi un routing più flessibile e un'ottimizzazione OTN. Inoltre, a livello WDM, un posizionamento ottimizzato degli amplificatori ottici (OA) nelle reti metropolitane può anche offrire opportunità di minimizzazione dei costi. Gli OA vengono utilizzati per aumentare la portata del segnale nei sistemi WDM, consentendo l'aumento simultaneo di più lunghezze d'onda e riducendo la necessità di costose apparecchiature di rigenerazione. Il posizionamento ottimizzato degli OA può garantire un'architettura metropolitana ottica a basso costo rimuovendo gli OA non necessari, riducendo il consumo di energia e massimizzando la qualità del segnale. Il posizionamento corretto degli OA consente di compensare la perdita di segnale contrastando al contempo il rumore e gli effetti non lineari. Considerando sia il traffic grooming, sia i vincoli correlati a FON e il posizionamento OA, si ottiene una progettazione complessa e non banale del layer logico e fisico congiunto. Ad esempio, il posizionamento OA può influenzare pesantemente il costo del layer OTN in termini di schede di grooming a causa dell'interdipendenza del posizionamento OA e dei lightpath stabiliti (e del loro bit rate) nella soluzione di grooming. Questa tesi elabora nuovi approcci di ottimizzazione per la minimizzazione dei costi delle apparecchiature OTN a singolo strato e OTN-over-WDM multistrato, considerando diversi scenari di grooming del traffico (ad esempio, ottenuti utilizzando schede di grooming rispetto a muxponder), architettura dei nodi senza filtro (ad esempio, FON rispetto a PFON) e posizionamento OA ottimizzato e considerando la retrocompatibilità con le tecnologie a 10 Gbps. Gli approcci proposti sono testati su topologie di rete reali e matrici di traffico fornite dai nostri collaboratori industriali. Questa tesi esplora le ottimizzazioni a singolo strato e multistrato sfruttando diverse euristiche, meta-euristiche e metodi di programmazione lineare intera (ILP). Più precisamente, la tesi presenta quattro contributi principali, due incentrati sull' ottimizzazione a strato OTN e due incentrati sull'ottimizzazione OTN-over-WDM multistrato. Innanzitutto, proponiamo diversi approcci di ottimizzazione per minimizzare i costi delle apparecchiature OTN a strato OTN, decidendo al contempo la soluzione di grooming e routing più adatta. In secondo luogo, introduciamo una nuova apparecchiatura, chiamata muxponder, nel layer OTN, per esplorare il suo impatto sulla minimizzazione dei costi. In terzo luogo, proponiamo diverse tecniche multistrato per l' ottimizzazione congiunta dei costi delle apparecchiature nei layer OTN e WDM. Infine, esaminiamo l'impatto di diverse strutture di nodi, come FON, PFON e WSS, sull'ottimizzazione dei costi multistrato.
Multi-layer design with low-cost of equipment in OTN-over-WDM metro networks
ATTARPOUR, ARYANAZ
2024/2025
Abstract
As the interconnecting segment between access and the core networks, optical metro networks must accommodate ever-increasing and heterogeneous traffic demands, from low-rate demands (e.g., 1 Gbps to 10 Gbps) fueled by recent deployments of Fiber-to-the-Home (FTTH), to large demands supporting interconnection of the data centers (that can require up to 400 Gbps). Moreover, new services and applications are being introduced following the emergence of new technologies, such as smart cities, artificial intelligence, etc, that also contribute to the increase in the number of customers and requested capacity. Therefore, novel solutions to increase the capacity of optical metro networks are constantly sought after by vendors and operators worldwide. Such solutions must support recent coherent transmission technologies to support high transmission rates, but also ensure compatibility with the current metro transmission technologies, as e.g., legacy 10G non-coherent transmission technologies. Accordingly, new and scalable optimization approaches must be devised to achieve low-cost deployment of equipment and to efficiently allocate resources in the optical metro network. Different technologies can be considered as low-cost enablers at various layers of the optical metro networks. In this work, we focus on the combined application of three such enablers: traffic grooming, filterless node architectures, and optimized optical-amplifier placement. Traffic grooming consists in enabling the aggregation of low-rate clients (e.g., 1G and 10G) to higher-rate ones (e.g., 10G and 100/200/400G) using electronic grooming equipment at the Optical Transport Network (OTN) layer. Traffic grooming allows for reducing the number of OTN lines at the cost of increasing the number of grooming boards. The OTN layer can support different grooming boards that can enable traffic grooming in different patterns. As an example, the hierarchy of OTN boards can enable hierarchical traffic grooming, allowing various choices of granular traffic grooming or muxponders that can enable less granular traffic grooming but allow to reach higher rates and do not need separate grooming matrices as traditional boards, hence enforcing new types of constraints in the grooming problem. At the Wavelength Division Multiplexing (WDM) layer, filterless node architecture has been demonstrated to provide significant cost savings. In Filterless Optical Networks (FONs), the costly Wavelength Selective Switches (WSS) used in Wavelength-Switched Optical Networks (WSON) are substituted with inexpensive and unadorned passive equipment such as splitters and combiners. This replacement results in lower deployment costs and less power consumption. However, it causes spectrum waste since the Amplified Spontaneous Emission (ASE) noise generated by optical amplifiers circulates in the network even after reaching the destination due to a lack of WSS filtering, making the reuse of spectrum impossible. In FONs, the interconnection of the nodes is fixed and does not permit a flexible interconnection in case of dynamic traffic or traffic change, resulting in more constrained routing than WSON, which provides all-to-all node interconnections by leveraging WSSes. The significant limitation to the routing flexibility can be addressed using the so-called Programmable FONs (PFONs) that include in the node architecture some spatial switching element to enable at least fiber switching capabilities. PFONs provide an alternative to FONs, as they can still enable the low-cost benefits of the FON while partially limiting its drawbacks, as spatial switches can at least provide reconfigurable fiber-level interconnection of the nodes, hence achieving more flexible routing and OTN optimization. Moreover, at the WDM layer, an optimized placement of Optical Amplifiers (OAs) in metro networks can also offer opportunities for cost minimization. OAs are used to increase signal reach in WDM systems, enabling the simultaneous boost of multiple wavelengths and reducing the need for costly regeneration equipment. The optimized placement of OAs can guarantee low-cost optical metro architecture by removing unnecessary OAs, reducing power consumption, and maximizing signal quality. Proper placement of OAs allows to compensate signal loss while counteracting noise and nonlinear effects. Considering both traffic grooming, FON-related constraints and OA placement results in a complex and non-trivial joint logical and physical layer design. For example, OA placement can heavily influence the cost of the OTN layer in terms of grooming boards due to the interdependence of the OA placement and established lightpaths (and their bit rate) in the grooming solution. This thesis devises novel optimization approaches for both single-layer OTN to multi-layer OTN-over-WDM equipment cost minimization, considering different scenarios of traffic grooming (e.g., obtained using grooming boards vs. muxponders), filterless node architecture (e.g., FON vs PFON), and optimized OA placement, and considering backward compatibility with 10Gbps technologies. The proposed approaches are tested on real network topologies and traffic matrices provided by our industrial collaborators. This thesis explores single-layer and multi-layer optimizations by leveraging different heuristics, meta-heuristics, and Integer Linear Programming (ILP) methods. More precisely, the thesis presents four main contributions, two focusing on OTN-layer optimization and two focusing on cross-layer OTN-over-WDM optimization. First, we propose different optimization approaches to minimize the cost of OTN equipment at the OTN layer while deciding on the best-fitted grooming and routing solution. Second, we introduce a new equipment, called a muxponder, into the OTN layer to explore its impact on cost minimization. Third, we propose different multi-layer techniques for joint equipment cost optimization at the OTN and WDM layers. Finally, we investigate the impact of different node structures, such as FON, PFON, and WSON, on the multi-layer cost optimization.| File | Dimensione | Formato | |
|---|---|---|---|
|
Attarpour_PhD_Tesi.pdf
non accessibile
Descrizione: PhD thesis
Dimensione
5.82 MB
Formato
Adobe PDF
|
5.82 MB | Adobe PDF | Visualizza/Apri |
I documenti in POLITesi sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
https://hdl.handle.net/10589/241357