Metro-aggregation networks have unique requirements: capacity and cost targets in between those of the access and metro-core networks they interconnect, and survivability to (at least) single failures. Meeting these requirements can be achieved via low-cost pluggable transceivers and cost-effective optical nodes, deployed over a horseshoe topology. By maintaining connectivity between each (leaf) node and the hub nodes at the extremes of the horseshoe, survivability to single link or node failures is guaranteed. However, in certain countries/regions, frequent power outages or enforcement of load shedding can make these networks temporarily vulnerable to single failures. Recently, it has been shown that resiliency against these events can be improved by adopting a filterless node architecture and minimizing the utilization of active devices in the express path of the optical signals. This paper extends the seminal work by proposing extensions to the original design framework that allow it to cover a larger number of network scenarios. Moreover, it reports the outcome of a comprehensive set of network simulations that provide insight into the impact that the coupler types used to realize the optical nodes have on both the number of required optical amplifiers and the vulnerability to local power outages and load shedding.
Quantum technologies promise to boost ICT further by extending its limits, even by promising more effective and secure communications due to the benefits of quantum physics. This leads to Quantum Key Distribution (QKD), a cryptographic primitive for securely exchanging a secret key between distant endpoints, leveraging an insecure communication channel due to potential eavesdroppers. The literature on QKD started with the BB84 protocol proposed in 1984 and continued with other protocols, each with its pros and cons. Still, they have been chiefly designed for static scenarios and do not consider the needs of dynamic networks. This paper presents a QKD solution for dynamic networks combining adaptive key management techniques and fault-tolerant routing, ensuring high security and efficiency even in variable network topologies. Using MDI-QKD (Measurement Device Independent QKD) and simulations with NetSquid, we provide a preliminary assessment of this solution's robustness against dynamic attacks and the improvement in quantum resource management. The work also compares the protocol's effectiveness with existing solutions, highlighting the advantages and potential practical applications.
Markov models are widely used to analyze complex stochastic systems, particularly in modeling of server cluster resilience, where they provide valuable insights into key performance metrics such as unavailability and blocking probability. However, their applicability to large-scale systems is often constrained by the state space explosion problem, leading researchers to employ decomposition-based approximation techniques to improve computational efficiency. Some commonly used approaches involve nearly-completely decomposable (NCD) Markov models, where processes operating at significantly different time scales are treated separately. This work critically examines the assumption that a large difference in time scales between system components guarantees a small approximation error. Using two examples from the context of server cluster resilience, we demonstrate that simplifications in NCD Markov models can lead to substantial inaccuracies. While decomposition-based methods offer large computational savings, their accuracy depends on specific structural conditions, rather than just the separation of time scales. Our findings emphasize the importance of carefully validating approximation assumptions and model structures to ensure reliable performance evaluation in large-scale stochastic systems.
Attack detection plays a crucial role in managing the security of communication networks. Thus, advanced methods have been developed to tackle this problem. However, the need for more flexible and adaptable detection systems remains, able to generalize across various attack scenarios and retain model performance over time. In this paper, we address these challenges by proposing two versions of a continuously updating attack detector in the form of binary classifier. Experimental evaluation on diverse datasets and various base classifiers confirms the effectiveness of our methodology.
In this paper, we revisit adaptive proactive protection schemes for backbone networks to guarantee high service availability. We aim to predict available - spare - link capacities and to use them as protection bandwidth to meet the service availability requirements of each connection. With a reasonable discretization of time, the advent of highly scalable capacity-and disaster risk-aware routing algorithms enables minimizing the lookahead at 2. In other words, we claim the spare bandwidth has to be predicted for only 2 time slots ahead. While minimizing the lookahead plays a key role in achieving better predictions, foreseeing the network bandwidth usage proves to remain a notoriously difficult task: traffic patterns are highly dynamic and are influenced by factors such as user behavior, daily, weekly, or yearly periodicity, and even unforeseeable external events. After discussing the protection scheme design, this study compares the performance of two time series forecasting methods for the prediction of network traffic in the short term: Autoregressive Integrated Moving Average (ARIMA) models and Temporal Convolutional Networks (TCN). These models are assessed on real traffic data obtained from the Energy Sciences Network (ESnet), a high-speed, international backbone network.
