Towards C+L band elastic optical networks: a solution to overcome the opticalfiber capacity crunch problem

Abstract

Over the past decade, there has been an exponential rise of high-data-rate, ultra-low-latency, and bandwidth-hungry applications at the client’s end, which has increased data traffic many-fold. Recent developments in 5G and cloud-based services have driven the compound annual growth rate (CAGR) of global Internet protocol (IP) traffic by nearly 30%. Hence, in the near future, per-month IP traffic growth will reach the zetta-bytes era. The COVID-19 pandemic has also catalyzed the usage of online services, leading to a 50% increase in data traffic compared to the pre-pandemic scenario. Recent forecasts indicate that, for the next ten years, the deployed infrastructure of the optical fiber backbone network will need to be upgraded to support this exponential growth of IP traffic to avoid the fiber-capacity-crunch problem. This has compelled network operators to seek out new strategies for enhancing network capacity while increasing minimum capital expenditure (CapEx). The use of multiband (MB) and multifiber technologies appears to be a promising alternative immediate solution in recent studies while having its own trade-o↵ to address this fiber capacity crunch problem and enhance the overall network capacity. MB technology considers the exploitation of the deployed standard single-mode fiber while enabling transmission on other bands (such as O, E, S, and L) along with the traditional C-band (1530 - 1565 nm). Although the multiband technology can use the full capacity of the existing deployed infrastructure, the challenges come from nonlinear impairments such as interchannel stimulated Raman scattering (ISRS). On the contrary, the multifiber-based solution focuses on the enlightenment of available dark fibers or the addition of additional parallel fibers. However, the CapEx for additional fiber resource deployment can play a crucial role in this context. The research in this thesis has focused on addressing the following fundamental questions in this context: • What is the cost-effective solution between MB and multifiber transmission? We justify this by performing techno-economic comparisons between these two solutions. As the L-band wavelengths provide the minimum attenuation after the C-band and as the L-band amplification appears as the most viable solution using the available commercial amplifiers, this thesis entirely focuses on C+L band transmission for analyzing the multiband scenarios. • Do the conventional resource allocation and spectrum management policies also work well for C+L band systems while considering the impact of physical layer impairments? • What will be the effective strategy to upgrade the existing C-band network towards the C+L band solution? • How can network survivability be ensured in the C+L band environment for geographically diverse networks? In the first part of this thesis, we have considered the C+L band physical layer model for estimating the quality of transmission (QoT) in multi-hop links while considering nonlinear interference (NLI) due to ISRS and other linear impairments (such as amplified spontaneous emission (ASE) noise). Consequently, we have made the techno-economic comparisons between the C+L band optical network with the multifiber C band scenario in the context of geographically diverse networks. Our results capture the variation of cost-per-bit with network traffic growth and fiber leasing cost. The numerical result indicates that the transmission over C+L bands is cost-effective compared to the multifiber C band scenario, particularly for large link-based geographies while considering the fiber leasing scenario. However, for smaller geographies, C+L band transmission becomes advantageous only if the fiber lease cost is high. As an alternate solution to fiber leasing, cable deployment scenarios are also considered separately for techno-economic comparison. The reported result shows that the C+L band system can postpone the need for extra cable deployment compared to the C band system and thereby minimizes the cost-per-bit in the long run. Furthermore, the comparison between optical cable deployment and fiber leasing is captured to minimize cost-per-bit while considering an operator’s domain knowledge about their network. After showcasing the benefits of C+L band transmission on cost-per-bit minimization, the next part of this thesis is focused on efficient spectrum management policies for the C+L band network to enhance the overall network capacity. Hitless spectrum defragmentation is mainly considered for the efficient utilization of spectral resources. The proposed work provides insights for network operators to develop the quality of service (QoS) maintenance strategy while doing spectrum defragmentation in the C + L bands. The proposed scheme prioritizes the minimization of the fragmentation index while maintaining the quality of transmission (QoT) for two different defragmentation algorithms, namely, nonlinear-impairment (NLI)-aware defragmentation (NAD) and NLI-unaware defragmentation (NUD). We leverage machine learning (ML) techniques for QoT estimation of ongoing lightpaths during spectrum reprovisioning. The optical signal-to-noise ratio (OSNR) of a lightpath is predicted for each choice of spectrum reprovisioning, which helps to monitor the effect of defragmentation on the quality of active lightpaths (in terms of assigned modulation format). Numerical results show that, compared to a baseline algorithm (NUD), the proposed NAD algorithm provides significant capacity increment for smaller and as well as larger networks. As the C+L band network is highly vulnerable to NLI, appropriate channel allocation during lightpath provisioning also becomes crucial for boosting the achievable capacity of the overall network. As a consequence, we have developed an efficient spectrum allocation strategy for C+L band networks. Unlike conventional schemes, the proposed scheme takes the effect of physical layer impairments (PLI) before the choice of the spectrum during resource allocation to reduce the likelihood of blocking. An algorithm under this scheme, namely, OSNR adaptive first–last-fit (OA-FLF), is proposed while leveraging the heterogeneity of the C+L band elastic optical network. The proposed algorithm selectively chooses the available channels among C and L bands to achieve the maximum network capacity in the long run. However, as the network evolves from the beginning-of-life (BoL) or end-of-life (EoL) situation, the analytical method of OSNR estimation under this OA-FLF algorithm becomes computationally intensive. We have leveraged the computational advantage of ML techniques to resolve this issue and utilized a deep neural network (DNN) model to predict the OSNR of all lightpaths during provisioning. Reported results show that, compared to the baseline algorithms, the proposed OA-FLF algorithm can provide gain in terms of traffic admissibility for smaller and as well as larger networks. As the existing C-band-based network can’t adopt these emerging technologies instantaneously, strategic planning needs to be done to upgrade the existing network infrastructure. Hence, the later part of this thesis focuses on efficient network upgrade strategies. First, the advantage of resource re-provisioning via selective movement of lightpaths from C-band to L-band is explored before upgrading the C-band network to C+L-bands. Later on, a novel strategy, named C to C+L Upgrade (CLU), is proposed to upgrade links from C to C+L bands gradually. We develop a recurrent neural network (RNN)-based model to efficiently predict links for the upgrade based on network state and spectrum utilization to reduce blocking and upgrade costs. Our results show that CLU outperforms baseline strategies (which do not employ predictive decisions) by upgrading fewer links at appropriate times. As the parallel multifiber C-band-based solution is beneficial for the network operators in the presence of their own dark-fibers availability, the later section of this thesis also explored the strategic network upgrade methodologies in this context. The advantage of adaptive margin allocation on network upgrades while effectively utilizing the monitoring data from optical performance monitoring equipment is shown for multifiber-based upgrade scenarios. The proposed approach considers periodic feedback from the network to gain domain knowledge and prioritizes adequate margin allocation before network upgrade initiation. Reported results show that usage of domain knowledge-assisted adaptive margin can postpone the requirement of the network upgrade, enhance the spectrum efficiency, and minimize cost-per-bit in the long run. Although operations over the C+L band can be an immediate and cost-effective solution for minimizing network upgrade costs, the impact on overall network reliability due to component failures needs to be considered for comprehensively assessing the true potential of the C+L band solution. Hence, the last part of this thesis is focused on the network survivability for C+L band networks while considering geographically diverse networks. In this context, we consider only single-band (either C or L band) inline amplifier failure scenarios. The provisioning of the backup lightpaths is prioritized over the same route as primary lightpaths using the alternate band. If the spectrum is unavailable in the primary routes, alternate routes are explored for backup path provisioning. Our approach measures the overall protection space of the network and the quality of the allocated lightpaths in geographically diverse networks. As a final step, we show the effect of the required Fill Margin (FM) on the achievable protection space and reliability of the network. To summarize, the research in this thesis is entirely focused on the C+L band optical networks while considering the impact of the physical layer impairments. Numerous significant aspects, such as techno-economic comparisons, cost-per-bit minimization, efficient spectrum resource management, periodic network upgrades, and reliability issues, have been explored in various parts of this thesis. The research output on these perspectives will be truly essential for network operators and vendors for practically enabling C+L band transmission in the core optical network.