A path-precomputation scheme is proposed to alleviate ATP delays in optical satellite networks. By forecasting requests on active links, it achieves up to 33.16% delay reduction, for 7.05× increase in runtime over conventional real-time computation.

The reduction in satellite launch costs has made the deployment of a large number of low Earth orbit (LEO) satellites possible. These satellites establish laser inter-satellite links (ISLs) to provide global coverage and high-speed data transmission. However, due to the mobility nature of satellite platforms, the wireless laser link is not as stable as the terrestrial fiber link. Laser ISLs are subject to predictable failures and random failures, both of which frequently cause connectivity handovers and transmission interruptions. To manage such challenges, this work tries to optimize the connectivity handover process with the aim of reducing the total interruption duration. We first define a dynamic satellite network model by introducing the time-to-disconnection risk for each ISL. Based on the network model, we design a proactive connectivity handover algorithm to determine the switching time and path for each handover operation, instead of waiting till the connectivity is interrupted. Simulation results demonstrate that the proposed proactive handover algorithm can effectively reduce the interruption rate by 41.37% and improve the maximum reliable transmission time by up to 31.89% compared to the baseline SP algorithm.

This paper presents a controller selection algorithm based on maximizing node betweenness centrality (MNB), which heuristically selects controller locations and improves control efficiency. Simulation results show that MNB reduces node configuration operations by 4.27% compared with random selection in a 100-node network, and improves routing success probability by 34.3% compared with centralized control in a 24-node topology.

A key aware dual-granularity channel adjustment algorithm is proposed for hybrid quantum- classical optical networks. It improves the success probability of security requests by 74.16% over the baseline.

We propose a pre-load-balancing (PLB) algorithm against satellite-ground link(SGL) attenuation to proactively mitigate potential service interruption. The simulation result shows that PLB can reduce 53.4% service interruption and 39.81% latency jitter.

This paper presents a resource-efficient multipath protection scheme based on virtual concatenation between ground stations in optical satellite networks. Compared with the Remove-Find algorithm, the blocking rate is reduced by 14.9%. © 2024 The Author(s)

This paper identifies key topics related to networking technologies in quantum key distribution networks (QKDNs) and evaluates ongoing standardization efforts. It also discusses the standardization trend toward next-generation quantum networks. © 2025 The Authors

AI training, e.g., large language models (LLMs), has emerged as one of the most critical workloads in modern data centers. The rapid growth of model sizes and training datasets has driven the need to scale up training clusters. While optical circuit switches (OCSs) offer cost and latency advantages for building large-scale machine learning (ML) clusters, their physical layer limitations, such as non-negligible reconfiguration latency, hinder their ability to handle the dynamic traffic patterns of ML workloads, leading to degraded network throughput and prolonged training time. In this paper, we propose Op tiML to optimize the interaction between ML workloads and reconfigurable optical topologies. OptiML integrates three key functionalities: (1) cluster partition, (2) spatial-temporal traffic optimization, and (3) on-demand OCS reconfiguration, which collectively optimize GPU allocation, traffic scheduling, and topology adaptation to maximize OCS resource utilization. We demonstrate OptiML's efficacy on a 6-GPU testbed, achieving iteration time improvements ranging from 1.12× to 1.19×. Large scale simulations further validate its performance superiority, reducing iteration time by up to 31% compared to state-of-the art ML training schedulers.

The explosive growth of data generated by Internet of Things applications has driven large-scale distributed machine learning (DML) on IoT cloud and edge cloud platforms. DML workloads in modern AI clusters generate a large amount of bursty and iterative communication traffic, making communication a major bottleneck affecting DML job completion. Therefore, flow scheduling is a critical approach for optimizing DML. In practice, AI clusters need to handle jobs with different latency requirements simultaneously, including Service Level Objective (SLO) jobs with strict deadlines and best-effort jobs without deadlines. However, existing DML flow schedulers ignore deadline requirements, resulting in low deadline satisfaction rates. We observe that communication contention among different jobs is a key factor causing jobs to fail to meet deadlines. In this paper, we propose Allot, a deadline-aware flow scheduler that mitigates the impact of inter-job communication contention on the deadline satisfaction rate of jobs. Allot leverages the periodicity and predictability of DML workloads to infer the job completion time and the spatial distribution of traffic. Based on this information, Allot performs admission control and adopts the Contention-Penalty Earliest Deadline First (CP-EDF) strategy to prioritize flows. We further introduce an efficient priority compression algorithm to adapt to limited priority levels on practical NICs and switches. Simulation results show that Allot can improve the job deadline satisfaction rate by an average of 1.13× to 5.03× compared to existing solutions.

Laser communication, with advantages in transmission distance, capacity, confidentiality, and anti-interference, has become a key development direction for future satellite communications, especially inter-satellite communications. However, satellite user service demands are unevenly distributed, with service hotspots mostly in the northern hemisphere, leading to traffic aggregation in developed areas. This issue is more prominent in low Earth orbit (LEO) constellations due to the large number of satellites and small ground service coverage. Therefore, effective routing algorithms are needed to alleviate congestion caused by traffic aggregation. In this paper, an advance planning-path conflict avoidance (AP-PCA) algorithm based on topology switching and traffic distribution awareness is proposed to avoid path conflict, considering the impact of dynamic topology switching and the performance difference of satellite–ground links. The influence of other original source–destination node pairs on critical links in the calculation of link weights to avoid conflicts in path selection is also considered. Compared with the three benchmark algorithms, the simulation results verify that our proposed solution reduces the blocking probability of traffic requests while ensuring the transmission delay and bandwidth utilization.

Quantum communication is envisioned as a foundational technology for future secure communication. Owing to the stable transmission properties of optical fibre, quantum communication is likely to be implemented over fibre infrastructures, currently with quantum key distribution (QKD) being the most representative protocol for deployment. The optical infrastructure is dominated by data services, where limited network resources are allocated for service provisioning. To cost-effectively scale QKD services to support multi-point users’ secured communication, it remains a critical challenge to share the existing optical networks’ capacity without disrupting legacy data services. At the same time, the variant distribution of data traffic leads to dynamic interference on the traversed QKD systems. In this paper, we focus on the QKD-integrated optical networks scenario, and investigate the network capacity maximization solutions. The network resources are shared by QKD and data services, where quadratic programming is utilised to optimise the service provisioning while minimising its interference on traversed quantum channels. Simulation results demonstrate that the outcome solution increases the average secret key rate in a 9-node mesh network from 0.38 Mbit/s to 6.51 Mbit/s, comparing with the first-fit benchmark. This outcome is achieved without blocking any data service request.

Optical satellite networks, supported by optical inter-satellite links (OISLs), provide reliable and low-latency optical connectivity. However, periodic and predictable sun outage events significantly compromise OISL availability, leading to frequent OISL interruptions and reduced network reliability. Existing routing algorithms often overlook the regularity of sun outage-induced interrupts and their differentiated impacts on services, resulting in degraded service performance. To address this challenge, this paper proposes a sun outage-enhanced time discretization OISL model and introduces a sun outage link-aware routing (SOLR) algorithm. By incorporating joint awareness of sun outage patterns and service requirements, SOLR employs an adaptive optimization mechanism to dynamically adjust routing decisions within temporal windows. Experimental results demonstrate that SOLR extends stable path durations by 39.9%, reduces interruption rates by 28.5%, and decreases blocking rates by 36.4%, significantly outperforming link-state-based routing algorithms. By effectively mitigating the impact of sun outages, SOLR ensures continuous optical service connections. This interruption-tolerant framework bridges network modeling and service provisioning, offering a robust solution for mission-critical service in optical satellite networks.

The deployment of large-scale satellite networks demands high-capacity and stable inter-satellite communication links, thereby driving the adoption of coherent optical satellite communication (COSC). However, relative satellite motion introduces Doppler shifts, which severely degrade link performance. To address this challenge, we discover the periodic correlation between Doppler shifts and the Gardner timing error detector (TED). Based on this correlation, we propose a novel frequency-offset estimation (FOE) algorithm that estimates frequency offset by computing the Gardner TED gain. The proposed algorithm surpasses the conventional FOE’s half-baud-rate limitation and maintains high estimation accuracy under strong noise conditions. We conduct 25-Gbaud DP-QPSK transmission experiments to evaluate the performance of the proposed FOE and to further examine its role in Doppler-shift compensation. The experimental results demonstrate that the proposed algorithm achieves a Doppler-shift estimation range 1.9 times that of conventional FOE algorithms. In addition, providing noise-robust Doppler-shift estimation improves receiver sensitivity by 0.6 dB and ensures successful signal demodulation even at an OSNR of 10 dB, which is where conventional FOE algorithms fail.

In this study, we aim to further advance solutions to critical challenges in physical layer secure key generation and distribution (PLSKGD) for optical fiber communication, particularly the limitations of traditional information reconciliation (IR) protocols in mitigating initial key inconsistencies caused by noise and interference. We introduce two innovative approaches: a check node correlation coefficient-based early termination (CNCC-ET) scheme and an elastic reconfigurable (ER) scheme to enhance error correction performance and accelerate IR. The CNCC-ET scheme dynamically monitors the correlation coefficient of messages between check and variable nodes, enabling early termination of iterations when decoding is likely to fail, thus significantly reducing decoding complexity. In high bit error rate (BER) regions, this scheme achieves up to a 51.8% reduction in average iteration counts while maintaining low frame error rates (FER). Concurrently, our ER scheme adapts to varying BER conditions by dynamically adjusting the ratio of information to check bits, achieving a remarkable 90.26% FER reduction in high-BER scenarios. Hardware implementation on a Xilinx Virtex-7 FPGA validates the feasibility of these methods, demonstrating a 47.2% reduction in IR time overhead and significantly improved error correction capability. This study advances the development of more secure and efficient optical fiber communication systems, with potential applications extending to other physical layer security contexts.

This paper proposes a turbulence-induced perturbation (TIP) approach to address the security degradation of quantum noise stream cipher (QNSC) systems under high transmit power and introduces a variable-order quantum noise stream cipher (VO-QNSC) scheme to further enhance transmission performance. The TIP approach incorporates turbulence-induced perturbation to strengthen the physical-layer security of QNSC in high-power scenarios, while the VO-QNSC scheme significantly improves system performance without increasing algorithmic complexity or redundancy, making it suitable for deployment on satellite terminals with limited computational resources. Simulation results show that, after introducing TIP, the system detection failure probability (DFP) can exceed 99.97% and number of masked signals (NMS) is improved to the order of 103, which effectively enhances the anti-eavesdropping ability of the system. For poor channel conditions, VO-QNSC can improve the receiver sensitivity by up to approximately 0.5 dB, which can meet the requirements of communication security and transmission performance in complex environments.

Quantum entanglement is one of the remarkable properties of the quantum world and can provide information-theoretic security. Although most current implementations are limited to two communication parties, the introduction of entanglement distribution networks offers the prospect of extending the advantages of entanglement to involve more than two users. However, simultaneous establishment of entanglement among numerous users still faces some challenging issues, such as the finite wavelength resources, which limit the scale and efficiency of entanglement distribution networks. Combined with subnet division, we propose the topological rotation symmetry-based wavelength allocation (TRS-WA) scheme for entanglement distribution networks, reducing the number of wavelength channels required for N users from the order of O(N) to O(root N). In order to improve the network mean effective entanglement probability (MEEP), we propose a hybrid multielement-based wavelength multiplexing (HM-WM) scheme, including the mixed use of the beam splitters and optical switches. In addition, the multi-objective linear programming model is designed to solve the problem of how to perform the wavelength allocation and wavelength multiplexing for a known size network. Simulation results indicate that the reduction in the number of wavelength channel pairs required by the TRS-WA scheme relative to the benchmark can reach 99% with 150 users.

Cell–free massive multiple–input multiple–output (CF–mMIMO) has attracted significant research interest by leveraging coherent joint transmission (CJT) for achieving excellent performance. However, this performance is degraded by phase offsets arising from imperfect radio frequency (RF) chains and asynchronous reception. To significantly recover CJT performance with low calibration feedback overhead, we propose a multi-reference calibration method. Specifically, we first propose a multi-reference phase offset alignment mechanism that designates multiple service access points (APs) as reference APs, thereby eliminating feedback for self-referencing service APs and aligning phase offsets to a bounded range. Leveraging this mechanism, we formulate a reference AP selection optimization problem to balance CJT performance recovery and feedback overhead, and develop a matching-based algorithm to solve it. Simulation results show that compared to conventional single-reference methods, our method reduces feedback overhead by 45.6% while incurring only a 2.3% performance loss.

This paper investigates cost-efficient task scheduling in computing power network with non-uniform node and link security capabilities. We propose a deep reinforcement learning approach that ensures secure computing and transmission while reducing overhead by 9%.