As guaranteed reliable experience (GRE) is one of the features of fifth-generation fixed networks (F5G), high-reliability optical transport networks (OTNs) have become one of the key technologies supporting this feature. Unfortunately, current OTN protection methods often provide fixed bandwidth for protection of 1 Gbps or more, which leads to resource wastage. Fine grain OTN (fgOTN) is an extension of existing OTN, which supports hitless bandwidth adjustment and uses 10 Mbps time slot isolation. The application of fgOTN’s advantages to network protection can save resources. However, how much initial protection bandwidth is reserved for links to improve the service recovery success probability after faults is a key issue to be studied. If the initially reserved protection bandwidth is too much, that may waste precious bandwidth resources and fail to recover other services. If the initially reserved protection bandwidth is too small, the controller needs to adjust the bandwidth frequently to meet service requirements, which puts tremendous pressure on network management and control. This study proposes a maximum bandwidth segmented shared protection (MBSSP) scheme, which is based on optimized centralized and distributed collaboration network management architecture. The protection scheme includes two algorithms: (i) the resource reservation algorithm used before the fault occurs based on maximum bandwidth segmented shared protection and (ii) the protection switch algorithm used after the fault occurs based on bandwidth adjustment. Simulative results show that, in a 38-node topology, compared with minimum bandwidth dedicated protection (MBDP), MBSSP only sacrifices 0.8% of resource utilization but can reduce the bandwidth adjustment probability by 15.8% and improves the recovery success probability by 33.4%. Compared with end-to-end shared protection (E2ESP), MBSSP improves recovery success probability by 42.9% and saves resources by 16.7%, although it increases the bandwidth adjustment probability by 20%.
{"title":"Segmented protection scheme based on maximum bandwidth sharing in F5G","authors":"Wenhong Liu;Yongli Zhao;Yajie Li;Xin Li;Sabidur Rahman;Jie Zhang","doi":"10.1364/JOCN.529958","DOIUrl":"https://doi.org/10.1364/JOCN.529958","url":null,"abstract":"As guaranteed reliable experience (GRE) is one of the features of fifth-generation fixed networks (F5G), high-reliability optical transport networks (OTNs) have become one of the key technologies supporting this feature. Unfortunately, current OTN protection methods often provide fixed bandwidth for protection of 1 Gbps or more, which leads to resource wastage. Fine grain OTN (fgOTN) is an extension of existing OTN, which supports hitless bandwidth adjustment and uses 10 Mbps time slot isolation. The application of fgOTN’s advantages to network protection can save resources. However, how much initial protection bandwidth is reserved for links to improve the service recovery success probability after faults is a key issue to be studied. If the initially reserved protection bandwidth is too much, that may waste precious bandwidth resources and fail to recover other services. If the initially reserved protection bandwidth is too small, the controller needs to adjust the bandwidth frequently to meet service requirements, which puts tremendous pressure on network management and control. This study proposes a maximum bandwidth segmented shared protection (MBSSP) scheme, which is based on optimized centralized and distributed collaboration network management architecture. The protection scheme includes two algorithms: (i) the resource reservation algorithm used before the fault occurs based on maximum bandwidth segmented shared protection and (ii) the protection switch algorithm used after the fault occurs based on bandwidth adjustment. Simulative results show that, in a 38-node topology, compared with minimum bandwidth dedicated protection (MBDP), MBSSP only sacrifices 0.8% of resource utilization but can reduce the bandwidth adjustment probability by 15.8% and improves the recovery success probability by 33.4%. Compared with end-to-end shared protection (E2ESP), MBSSP improves recovery success probability by 42.9% and saves resources by 16.7%, although it increases the bandwidth adjustment probability by 20%.","PeriodicalId":50103,"journal":{"name":"Journal of Optical Communications and Networking","volume":"16 11","pages":"1145-1158"},"PeriodicalIF":4.0,"publicationDate":"2024-10-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142524177","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
We propose an innovative optimization framework using a multi-objective genetic algorithm to simultaneously optimize the launch power profile and design Raman amplifiers. Its flexibility allows us to find better solutions and reduce the number of Raman pumps. Moreover, we utilize the framework to compare the potential of four multi-band transmission systems leveraging hybrid fiber amplification. Simulation results highlight that complementing a C + L-band system with the S-band leads to higher total system capacity than using the E-band or interleaving data channels and Raman pumps.
我们提出了一种创新的优化框架，利用多目标遗传算法同时优化发射功率曲线和设计拉曼放大器。它的灵活性使我们能够找到更好的解决方案，并减少拉曼泵的数量。此外，我们还利用该框架比较了四种利用混合光纤放大的多波段传输系统的潜力。仿真结果表明，与使用 E 波段或交错数据通道和拉曼泵相比，使用 S 波段对 C + L 波段系统进行补充可提高系统总容量。
{"title":"Raman amplifier design and launch power optimization in multi-band optical systems","authors":"Andre Souza;Nelson Costa;Joao Pedro;Joao Pires","doi":"10.1364/JOCN.534006","DOIUrl":"https://doi.org/10.1364/JOCN.534006","url":null,"abstract":"We propose an innovative optimization framework using a multi-objective genetic algorithm to simultaneously optimize the launch power profile and design Raman amplifiers. Its flexibility allows us to find better solutions and reduce the number of Raman pumps. Moreover, we utilize the framework to compare the potential of four multi-band transmission systems leveraging hybrid fiber amplification. Simulation results highlight that complementing a C + L-band system with the S-band leads to higher total system capacity than using the E-band or interleaving data channels and Raman pumps.","PeriodicalId":50103,"journal":{"name":"Journal of Optical Communications and Networking","volume":"17 1","pages":"A13-A22"},"PeriodicalIF":4.0,"publicationDate":"2024-10-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142518227","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Jun-ichi Kani;Takahiro Suzuki;Yasutaka Kimura;Shin Kaneko;Sang-Yuep Kim;Tomoaki Yoshida
Future access and metro networks are expected to provide advanced broadband services and the evolution of mobile x-haul in a flexible manner. This paper first reviews the progress and challenges of disaggregation and virtualization technologies to meet this expectation with a focus on their application to optical access networks. Next, it describes future access and metro integrated networking in which disaggregation and virtualization technologies will play important roles.
{"title":"Disaggregation and virtualization for future access and metro networks [Invited Tutorial]","authors":"Jun-ichi Kani;Takahiro Suzuki;Yasutaka Kimura;Shin Kaneko;Sang-Yuep Kim;Tomoaki Yoshida","doi":"10.1364/JOCN.534303","DOIUrl":"https://doi.org/10.1364/JOCN.534303","url":null,"abstract":"Future access and metro networks are expected to provide advanced broadband services and the evolution of mobile x-haul in a flexible manner. This paper first reviews the progress and challenges of disaggregation and virtualization technologies to meet this expectation with a focus on their application to optical access networks. Next, it describes future access and metro integrated networking in which disaggregation and virtualization technologies will play important roles.","PeriodicalId":50103,"journal":{"name":"Journal of Optical Communications and Networking","volume":"17 1","pages":"A1-A12"},"PeriodicalIF":4.0,"publicationDate":"2024-10-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=10734083","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142517860","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
The increasing complexity and dynamicity of future optical networks will necessitate accurate, fast, and low-cost quality-of-transmission (QoT) estimation. Machine learning-based QoT estimation models have shown promise in ensuring the reliability and efficiency of optical networks. However, the data-driven nature of these models impedes their application in practical settings. To address the problem of limited data availability in the target domain, known as the few-shot learning problem, we propose a domain adversarial adaptation method that aligns the distributions of representations from different source and target domains by minimizing the domain discrepancy quantified by the approximate Wasserstein distance. We demonstrate the method’s effectiveness through a theoretical proof and two example adaptations, i.e., from simulation to experimental data and from experimental to real network data. Our method consistently outperforms commonly used artificial neural networks (ANNs) and more advanced transfer learning approaches for various target domain data sizes. More profoundly, we show two ways to further improve the prediction accuracy, i.e., incorporating unlabeled target domain data in the training stage and utilizing the learned representations after training to train a new ANN with a reweighting strategy. In the adaptation to actual field data, our model, trained with only eight labeled network data samples, outperforms an ANN trained with 300 samples, thus reducing the labeled target domain data burden by more than 97%. The proposed method’s adaptability and generalizability make it a promising solution for accurate QoT estimation with low data requirements in future intelligent optical networks.
{"title":"Domain adversarial adaptation framework for few-shot QoT estimation in optical networks","authors":"Zhuojun Cai;Qihang Wang;Yubin Deng;Peng Zhang;Gai Zhou;Yang Li;Faisal Nadeem Khan","doi":"10.1364/JOCN.530915","DOIUrl":"https://doi.org/10.1364/JOCN.530915","url":null,"abstract":"The increasing complexity and dynamicity of future optical networks will necessitate accurate, fast, and low-cost quality-of-transmission (QoT) estimation. Machine learning-based QoT estimation models have shown promise in ensuring the reliability and efficiency of optical networks. However, the data-driven nature of these models impedes their application in practical settings. To address the problem of limited data availability in the target domain, known as the few-shot learning problem, we propose a domain adversarial adaptation method that aligns the distributions of representations from different source and target domains by minimizing the domain discrepancy quantified by the approximate Wasserstein distance. We demonstrate the method’s effectiveness through a theoretical proof and two example adaptations, i.e., from simulation to experimental data and from experimental to real network data. Our method consistently outperforms commonly used artificial neural networks (ANNs) and more advanced transfer learning approaches for various target domain data sizes. More profoundly, we show two ways to further improve the prediction accuracy, i.e., incorporating unlabeled target domain data in the training stage and utilizing the learned representations after training to train a new ANN with a reweighting strategy. In the adaptation to actual field data, our model, trained with only eight labeled network data samples, outperforms an ANN trained with 300 samples, thus reducing the labeled target domain data burden by more than 97%. The proposed method’s adaptability and generalizability make it a promising solution for accurate QoT estimation with low data requirements in future intelligent optical networks.","PeriodicalId":50103,"journal":{"name":"Journal of Optical Communications and Networking","volume":"16 11","pages":"1133-1144"},"PeriodicalIF":4.0,"publicationDate":"2024-10-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142518221","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Jose Manuel Rivas-Moscoso;Farhad Arpanaei;Gabriel Otero Perez;Jose David Martinez Jimenez;Juan Pedro Fernandez-Palacios;Oscar Gonzalez de Dios;Luis Miguel Contreras;Alfonso Sanchez-Macian;Jose Alberto Hernandez;David Larrabeiti;Jesus Folgueira
In this paper, we introduce TEFNET24, a reference multi-layer hierarchical network topology that spans from access to core networks, specifically designed to meet the demands of beyond 5G and prepared for next-generation 6G communication systems. This topology, inspired by the actual network deployments of Telefónica in medium-sized countries (or large federal states) in Europe and America, integrates both IP and optical (DWDM) layers to provide a comprehensive framework for network design, optimization, and analysis. Our primary contribution is the development of an open-source benchmarking network, accessible to both researchers and industry professionals. This resource aims to facilitate the study and advancement of integrated IP and optical networks, allowing researchers to address key challenges such as traffic aggregation, latency reduction, cost efficiency, and support for advanced applications. We provide guidelines for utilizing this benchmark network, enabling users to evaluate and enhance their solutions for AI-driven network management, ultra-reliable low-latency communication, enhanced mobile broadband, and massive machine-type communication. By sharing this detailed and practical benchmarking network, we seek to foster innovation and collaboration within the optical network community, driving forward the capabilities and performance of future communication networks. A dataset with TEFNET24 details is provided.
在本文中，我们介绍了 TEFNET24，这是一种从接入网到核心网的参考多层分级网络拓扑结构，专为满足 5G 之后的需求而设计，并为下一代 6G 通信系统做好了准备。该拓扑受西班牙电信公司在欧洲和美洲中型国家（或大型联邦州）实际网络部署的启发，集成了 IP 层和光（DWDM）层，为网络设计、优化和分析提供了一个全面的框架。我们的主要贡献是开发了一个开源基准网络，供研究人员和行业专业人员使用。该资源旨在促进对集成 IP 和光网络的研究和发展，使研究人员能够应对流量聚合、降低延迟、成本效率和支持高级应用等关键挑战。我们提供了使用该基准网络的指南，使用户能够评估和改进其解决方案，以实现人工智能驱动的网络管理、超可靠的低延迟通信、增强型移动宽带和大规模机器型通信。通过共享这一详细而实用的基准网络，我们力求促进光网络社区内的创新与合作，推动未来通信网络的能力和性能向前发展。我们提供了包含 TEFNET24 详细信息的数据集。
{"title":"TEFNET24: reference packet optical network topology for edge to core transport","authors":"Jose Manuel Rivas-Moscoso;Farhad Arpanaei;Gabriel Otero Perez;Jose David Martinez Jimenez;Juan Pedro Fernandez-Palacios;Oscar Gonzalez de Dios;Luis Miguel Contreras;Alfonso Sanchez-Macian;Jose Alberto Hernandez;David Larrabeiti;Jesus Folgueira","doi":"10.1364/JOCN.533131","DOIUrl":"https://doi.org/10.1364/JOCN.533131","url":null,"abstract":"In this paper, we introduce TEFNET24, a reference multi-layer hierarchical network topology that spans from access to core networks, specifically designed to meet the demands of beyond 5G and prepared for next-generation 6G communication systems. This topology, inspired by the actual network deployments of Telefónica in medium-sized countries (or large federal states) in Europe and America, integrates both IP and optical (DWDM) layers to provide a comprehensive framework for network design, optimization, and analysis. Our primary contribution is the development of an open-source benchmarking network, accessible to both researchers and industry professionals. This resource aims to facilitate the study and advancement of integrated IP and optical networks, allowing researchers to address key challenges such as traffic aggregation, latency reduction, cost efficiency, and support for advanced applications. We provide guidelines for utilizing this benchmark network, enabling users to evaluate and enhance their solutions for AI-driven network management, ultra-reliable low-latency communication, enhanced mobile broadband, and massive machine-type communication. By sharing this detailed and practical benchmarking network, we seek to foster innovation and collaboration within the optical network community, driving forward the capabilities and performance of future communication networks. A dataset with TEFNET24 details is provided.","PeriodicalId":50103,"journal":{"name":"Journal of Optical Communications and Networking","volume":"16 11","pages":"G28-G39"},"PeriodicalIF":4.0,"publicationDate":"2024-10-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142518162","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Yue Pang;Min Zhang;Yanli Liu;Xiangbin Li;Yidi Wang;Yahang Huan;Zhuo Liu;Jin Li;Danshi Wang
The optical network encompasses numerous devices and links, generating a significant volume of logs. Analyzing these logs is significant for network optimization, failure diagnosis, and health monitoring. However, the large-scale and diverse formats of optical network logs present several challenges, including the high cost and difficulty of manual processing, insufficient semantic understanding in existing analysis methods, and the strict requirements for data security and privacy. Generative artificial intelligence (GAI) with powerful language understanding and generation capabilities has the potential to address these challenges. Large language models (LLMs) as a concrete realization of GAI are well-suited for analyzing DCI logs, replacing human experts and enhancing accuracy. Additionally, LLMs enable intelligent interactions with network administrators, automating tasks and improving operational efficiency. Moreover, fine-tuning with open-source LLMs protects data privacy and enhances log analysis accuracy. Therefore, we introduce LLMs and propose a log analysis method with instruction tuning using LLaMA2 for log parsing, anomaly detection and classification, anomaly analysis, and report generation. Real log data extracted from the field-deployed network was used to design and construct instruction tuning datasets. We utilized the dataset for instruction tuning and demonstrated and evaluated the effectiveness of the proposed scheme. The results indicate that this scheme improves the performance of log analysis tasks, especially a 14% improvement in exact match rate for log parsing, a 13% improvement in F1-score for anomaly detection and classification, and a 23% improvement in usability for anomaly analysis, compared with the best baselines.
光网络包含众多设备和链路，会产生大量日志。分析这些日志对网络优化、故障诊断和健康监控意义重大。然而，光网络日志规模庞大、格式多样，这给我们带来了诸多挑战，包括人工处理成本高、难度大，现有分析方法对语义的理解不足，以及对数据安全和隐私的严格要求。具有强大语言理解和生成能力的生成人工智能（GAI）有望应对这些挑战。大型语言模型（LLMs）作为 GAI 的具体实现形式，非常适合分析 DCI 日志，可替代人类专家并提高准确性。此外，LLM 还能与网络管理员进行智能互动，实现任务自动化并提高运行效率。此外，使用开源 LLM 进行微调可保护数据隐私并提高日志分析的准确性。因此，我们引入了 LLM，并提出了一种使用 LLaMA2 进行指令调整的日志分析方法，用于日志解析、异常检测和分类、异常分析以及报告生成。从现场部署的网络中提取的真实日志数据被用于设计和构建指令调整数据集。我们利用该数据集进行了指令调整，并演示和评估了建议方案的有效性。结果表明，与最佳基线相比，该方案提高了日志分析任务的性能，特别是日志解析的精确匹配率提高了 14%，异常检测和分类的 F1 分数提高了 13%，异常分析的可用性提高了 23%。
{"title":"Large language model-based optical network log analysis using LLaMA2 with instruction tuning","authors":"Yue Pang;Min Zhang;Yanli Liu;Xiangbin Li;Yidi Wang;Yahang Huan;Zhuo Liu;Jin Li;Danshi Wang","doi":"10.1364/JOCN.527874","DOIUrl":"https://doi.org/10.1364/JOCN.527874","url":null,"abstract":"The optical network encompasses numerous devices and links, generating a significant volume of logs. Analyzing these logs is significant for network optimization, failure diagnosis, and health monitoring. However, the large-scale and diverse formats of optical network logs present several challenges, including the high cost and difficulty of manual processing, insufficient semantic understanding in existing analysis methods, and the strict requirements for data security and privacy. Generative artificial intelligence (GAI) with powerful language understanding and generation capabilities has the potential to address these challenges. Large language models (LLMs) as a concrete realization of GAI are well-suited for analyzing DCI logs, replacing human experts and enhancing accuracy. Additionally, LLMs enable intelligent interactions with network administrators, automating tasks and improving operational efficiency. Moreover, fine-tuning with open-source LLMs protects data privacy and enhances log analysis accuracy. Therefore, we introduce LLMs and propose a log analysis method with instruction tuning using LLaMA2 for log parsing, anomaly detection and classification, anomaly analysis, and report generation. Real log data extracted from the field-deployed network was used to design and construct instruction tuning datasets. We utilized the dataset for instruction tuning and demonstrated and evaluated the effectiveness of the proposed scheme. The results indicate that this scheme improves the performance of log analysis tasks, especially a 14% improvement in exact match rate for log parsing, a 13% improvement in F1-score for anomaly detection and classification, and a 23% improvement in usability for anomaly analysis, compared with the best baselines.","PeriodicalId":50103,"journal":{"name":"Journal of Optical Communications and Networking","volume":"16 11","pages":"1116-1132"},"PeriodicalIF":4.0,"publicationDate":"2024-10-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142517897","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Some data center networks have already started to use optical circuit switching (OCS) with potential performance benefits, including high capacity, low latency, and energy efficiency. This paper addresses a switching network design to maximize the network radix, i.e., the number of terminals connected to the network under the condition that a specified number of identical switches with the size $N times N$ and the maximum admissible blocking probability are given. Previous work presented a two-stage twisted and folded Clos network (TF-Clos) with a blocking probability guarantee for OCS, which has a larger network radix than TF-Clos with a strict-sense non-blocking condition. Expanding the number of stages allows for enhancing the network radix. This paper proposes a model designing an OCS three-stage TF-Clos structure with a blocking probability guarantee to increase the network radix compared to the two-stage TF-Clos. We formulate the problem of obtaining the network configuration that maximizes the network radix as an optimization problem. We conduct an algorithm based on an exhaustive search to obtain a feasible solution satisfying the constraints of the optimization problem. This algorithm identifies the structure with the largest network radix in non-increasing order to avoid unnecessary searches. Numerical results show that the proposed model achieves a larger network radix than the two-stage model.
{"title":"Optical circuit switched three-stage twisted-folded Clos-network design model guaranteeing admissible blocking probability","authors":"Ryotaro Taniguchi;Takeru Inoue;Kazuya Anazawa;Eiji Oki","doi":"10.1364/JOCN.535282","DOIUrl":"https://doi.org/10.1364/JOCN.535282","url":null,"abstract":"Some data center networks have already started to use optical circuit switching (OCS) with potential performance benefits, including high capacity, low latency, and energy efficiency. This paper addresses a switching network design to maximize the network radix, i.e., the number of terminals connected to the network under the condition that a specified number of identical switches with the size \u0000<tex>$N times N$</tex>\u0000 and the maximum admissible blocking probability are given. Previous work presented a two-stage twisted and folded Clos network (TF-Clos) with a blocking probability guarantee for OCS, which has a larger network radix than TF-Clos with a strict-sense non-blocking condition. Expanding the number of stages allows for enhancing the network radix. This paper proposes a model designing an OCS three-stage TF-Clos structure with a blocking probability guarantee to increase the network radix compared to the two-stage TF-Clos. We formulate the problem of obtaining the network configuration that maximizes the network radix as an optimization problem. We conduct an algorithm based on an exhaustive search to obtain a feasible solution satisfying the constraints of the optimization problem. This algorithm identifies the structure with the largest network radix in non-increasing order to avoid unnecessary searches. Numerical results show that the proposed model achieves a larger network radix than the two-stage model.","PeriodicalId":50103,"journal":{"name":"Journal of Optical Communications and Networking","volume":"16 11","pages":"1104-1115"},"PeriodicalIF":4.0,"publicationDate":"2024-10-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142518161","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Yongxin Xu;Xiaokai Guan;Wenqing Jiang;Xudong Wang;Weisheng Hu;Lilin Yi
End-to-end learning allows communication systems to achieve optimal performance compared with conventional blockwise structure design. By modeling the channel with neural networks and training the transmitter and receiver on this differentiable channel, the whole system can be jointly optimized. However, in existing schemes, channel modeling methods, such as the generative adversarial network and long short-term memory network, have complex architectures and cannot track channel changes, leading to less effective end-to-end learning. Meanwhile, the complexity of neural networks deployed at the transmitter and receiver is too high for practical applications. In this work, we propose an efficient and low-complexity end-to-end deep learning framework and experimentally validate it on a 100G passive optical network. It uses a noise adaptation network to model channel response and noise distribution and employs offline pretraining and online tracking training to improve the efficiency and accuracy of channel modeling. For the transmitter, it consists of a pattern-dependent look-up table (PDLUT) based on a neural network (NN-PDLUT) with a single convolutional layer. Further, the receiver is also an NN with a single convolutional layer; thus, the end-to-end signal processing is extremely simple. The experimental results show that end-to-end learning improves the receiver sensitivity by 0.85 and 1.59 dB compared with receiver-only equalization based on Volterra nonlinear equalization (VNLE) and joint equalization based on a PDLUT and a feed-forward equalizer, respectively. Moreover, the number of multiply–accumulate operations consumed by the transmitter and receiver in the end-to-end learning scheme is reduced by 75.7% compared with VNLE-based receiver-only equalization.
{"title":"Low-complexity end-to-end deep learning framework for 100G-PON","authors":"Yongxin Xu;Xiaokai Guan;Wenqing Jiang;Xudong Wang;Weisheng Hu;Lilin Yi","doi":"10.1364/JOCN.532742","DOIUrl":"https://doi.org/10.1364/JOCN.532742","url":null,"abstract":"End-to-end learning allows communication systems to achieve optimal performance compared with conventional blockwise structure design. By modeling the channel with neural networks and training the transmitter and receiver on this differentiable channel, the whole system can be jointly optimized. However, in existing schemes, channel modeling methods, such as the generative adversarial network and long short-term memory network, have complex architectures and cannot track channel changes, leading to less effective end-to-end learning. Meanwhile, the complexity of neural networks deployed at the transmitter and receiver is too high for practical applications. In this work, we propose an efficient and low-complexity end-to-end deep learning framework and experimentally validate it on a 100G passive optical network. It uses a noise adaptation network to model channel response and noise distribution and employs offline pretraining and online tracking training to improve the efficiency and accuracy of channel modeling. For the transmitter, it consists of a pattern-dependent look-up table (PDLUT) based on a neural network (NN-PDLUT) with a single convolutional layer. Further, the receiver is also an NN with a single convolutional layer; thus, the end-to-end signal processing is extremely simple. The experimental results show that end-to-end learning improves the receiver sensitivity by 0.85 and 1.59 dB compared with receiver-only equalization based on Volterra nonlinear equalization (VNLE) and joint equalization based on a PDLUT and a feed-forward equalizer, respectively. Moreover, the number of multiply–accumulate operations consumed by the transmitter and receiver in the end-to-end learning scheme is reduced by 75.7% compared with VNLE-based receiver-only equalization.","PeriodicalId":50103,"journal":{"name":"Journal of Optical Communications and Networking","volume":"16 11","pages":"1093-1103"},"PeriodicalIF":4.0,"publicationDate":"2024-10-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142450921","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
As DWDM transmission offers enhanced wavelength/spectrum parallelism, the capacity of optical networks has been substantially increased. Due to the theoretical capacity limit of C-band transmission over single-mode fibers, research into new frequency bands and parallel fibers has become very active. However, the hardware scale of current optical cross-connect nodes will explode with greater wavelength/spectrum and spatial parallelism. Three optical node/network architectures are presented in this paper that take advantage of one or both of these parallelism technologies. These architectures will provide a baseline for cost-effective and bandwidth-abundant future optical networks based on massive parallelism.
由于 DWDM 传输提供了更强的波长/频谱并行性，光网络的容量得到了大幅提高。由于单模光纤 C 波段传输的理论容量有限，对新频段和并行光纤的研究变得非常活跃。然而，随着波长/频谱和空间并行性的提高，目前光交叉连接节点的硬件规模将出现爆炸式增长。本文介绍了三种利用一种或两种并行技术的光节点/网络架构。这些架构将为未来基于大规模并行性的高成本效益和带宽充裕的光网络提供基线。
{"title":"Optical networking that exploits massive wavelength/spectrum and spatial parallelisms","authors":"Hiroshi Hasegawa","doi":"10.1364/JOCN.532594","DOIUrl":"https://doi.org/10.1364/JOCN.532594","url":null,"abstract":"As DWDM transmission offers enhanced wavelength/spectrum parallelism, the capacity of optical networks has been substantially increased. Due to the theoretical capacity limit of C-band transmission over single-mode fibers, research into new frequency bands and parallel fibers has become very active. However, the hardware scale of current optical cross-connect nodes will explode with greater wavelength/spectrum and spatial parallelism. Three optical node/network architectures are presented in this paper that take advantage of one or both of these parallelism technologies. These architectures will provide a baseline for cost-effective and bandwidth-abundant future optical networks based on massive parallelism.","PeriodicalId":50103,"journal":{"name":"Journal of Optical Communications and Networking","volume":"16 11","pages":"H27-H39"},"PeriodicalIF":4.0,"publicationDate":"2024-10-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142443146","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
The lack of standards in the performance evaluation of new resource allocation algorithms in multicore fiber elastic optical networks (MCF-EONs) compromises the fairness when comparing them with the state of the art. This paper reviews the different transmission parameters, network parameters, performance metrics, and baselines used by the recent proposals to build a framework for future benchmarking of such algorithms according to the nature of the network operation, whether static or dynamic. This framework aims to provide standards regarding evaluation criteria, scenarios, and performance metrics, as well as recommendations concerning technology advances to promote methodology and reproducibility in further related studies.
{"title":"Benchmarking framework for resource allocation algorithms in multicore fiber elastic optical networks","authors":"Juan Pinto-Rios;Barbara Dumas Feris;Christofer Vasquez;Gabriel Saavedra;Danilo Borquez-Paredes;Nicolas Jara;Ricardo Olivares;Saquib Amjad;Ariel Leiva;Carmen Mas-Machuca","doi":"10.1364/JOCN.534257","DOIUrl":"https://doi.org/10.1364/JOCN.534257","url":null,"abstract":"The lack of standards in the performance evaluation of new resource allocation algorithms in multicore fiber elastic optical networks (MCF-EONs) compromises the fairness when comparing them with the state of the art. This paper reviews the different transmission parameters, network parameters, performance metrics, and baselines used by the recent proposals to build a framework for future benchmarking of such algorithms according to the nature of the network operation, whether static or dynamic. This framework aims to provide standards regarding evaluation criteria, scenarios, and performance metrics, as well as recommendations concerning technology advances to promote methodology and reproducibility in further related studies.","PeriodicalId":50103,"journal":{"name":"Journal of Optical Communications and Networking","volume":"16 11","pages":"G11-G27"},"PeriodicalIF":4.0,"publicationDate":"2024-10-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142438501","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：