本期目录

2022年 29卷 6期 刊出日期：2022-12-30



Special Topic: Optical Communication and Artificial Intelligence

Artificial intelligence for optical transport networks: architecture, application and challenges

Li Yajie, Zhang Jie

中国邮电高校学报(英文), 2022, 29 (6). doi： 10.19682/j.cnki.1005-8885.2022.1024

摘要 ( 479 ) PDF (4069 KB)( 59 )

参考文献 | 相关文章 | 计量指标

Optical network plays an important role in telecommunication networks, which supports high-capacity and long-distance transmission of Internet traffic. However, as the scaling and evolving of optical networks, it faces great challenges in terms of network operation, optimization and maintenance. Artificial intelligence ( AI ) has been proved to have superiority on addressing complex problems, by mimicking cognitive skills similar with human mind. In this paper, we provide a comprehensive investigation of AI applications in optical transport network. First, we give a general AI-based control architecture for optical transport networks. Then, we discuss several typical applications of AI model and algorithms in optical networks. Different use cases are considered, including network planning, quality of transmission ( QoT ) estimation, network reconfiguration, traffic prediction, failure management and so on. In addition, we also present some potential technical challenges for AI application in

optical network for the next years.

Reinforced virtual optical network embedding algorithm in EONs for edge computing

Zhu Ruijie, Li Gong, Wang Peisen, Zhang Wenchao

中国邮电高校学报(英文), 2022, 29 (6). doi： 10.19682/j.cnki.1005-8885.2022.1020

摘要 ( 286 ) PDF (3028 KB)( 39 )

参考文献 | 相关文章 | 计量指标

As the core technology of optical networks virtualization, virtual optical network embedding ( VONE) enables multiple virtual network requests to share substrate elastic optical network ( EON) resources simultaneously and hence has been applicated in edge computing scenarios. In this paper, we propose a reinforced virtual optical network embedding ( R-VONE ) algorithm based on deep reinforcement learning ( DRL) to optimize network embedding policies automatically. The network resource attributes are extracted as the environment state for model training, based on which DRL agent can deduce the node embedding probability. Experimental results indicate that R-VONE presents a significant advantage with lower blocking probability and higher resource utilization.

Genetic algorithm based dynamic combination of polyhedron structures against multiple link failures

Liu Yu, Li Xin, Huang Shanguo

中国邮电高校学报(英文), 2022, 29 (6). doi： 10.19682/j.cnki.1005-8885.2022.1021

摘要 ( 221 ) PDF (2337 KB)( 23 )

参考文献 | 相关文章 | 计量指标

Polyhedron protection realizes link protection by constructing a pre-assigned structure and allocates backup resources on a fixed polyhedron structure based on the maximum number of working resources. Taking into account both protection success rate and resource redundancy, this paper dynamically combines different polyhedron structures to allocate backup resources according to the link load, and proposes a genetic algorithm based dynamic combination of polyhedron structures (GA-DCPS) to reduce the resource consumption in the network while ensuring the protection success rate. GA-DCPS aims to minimize the consumption of wavelength resources, and uses the genetic strategy to find the polyhedron combination with the least redundancy to allocate backup resources while ensuring the success rate of service protection. Compared to using the fixed polyhedron structure with 1:m backup resource allocation, GA-DCPS can reduce resource redundancy by about 15% while ensuring complete protection against double-link failures.

Deep learning-based symbol detection algorithm in IMDD-OOFDM system

Zhang Huibin, Li Tianzhu, Liu Haojiang, Li Zhuotong

中国邮电高校学报(英文), 2022, 29 (6). doi： 10.19682/j.cnki.1005-8885.2022.1022

摘要 ( 272 ) PDF (4076 KB)( 35 )

参考文献 | 相关文章 | 计量指标

In the current research on intensity-modulation and direct-detection optical orthogonal frequency division multiplexing ( IMDD-OOFDM ) system, effective channel compensation is a key factor to improve system performance. In order to improve the efficiency of channel compensation, a deep learning-based symbol detection algorithm is proposed in this paper for IMDD-OOFDM system. Firstly, a high-speed data streams symbol synchronization algorithm based on a training sequence is used to ensure accurate symbol synchronization. Then the traditional channel estimation and channel compensation are replaced by an echo state network (ESN) to restore the transmitted signal. Finally, we collect the data from the system experiment and calculate the signal-to-noise ratio (SNR). The analysis of the SNR optimized by the ESN proves that the ESN-based symbol detection algorithm is effective in compensating nonlinear distortion.

Prediction-based dynamic routing intelligent algorithm in power optical communication network

Guo Xuerang, Li Feng, Zhu Bohan, Zhang Zhijun, Guo Qingrui, Yang Huiting

中国邮电高校学报(英文), 2022, 29 (6). doi： 10.19682/j.cnki.1005-8885.2022.1023

摘要 ( 278 ) PDF (1252 KB)( 51 )

参考文献 | 相关文章 | 计量指标

New energy power generation equipment has the characteristics of diurnal, perturbative, seasonal, and periodic power generation, which makes new power optical communication network ( POCN ) more dynamic and changeable. This is directly reflected in the dynamics of the link risk and service importance of the POCN. In this paper, aiming at the problem of the dynamic importance of service in POCN, and the resulting power optical communication network reliability decline problem, a new energy POCN dynamic routing intelligence algorithm based on service importance prediction is proposed. Based on the short-term power generation of new energy power station, the importance of each service and the risk degree of each link are predicted. Link weights are dynamically adjusted, and k-shortest path ( KSP) algorithm is used to calculate route results. When network resources are insufficient, low-importance services can give way to prevent a large number of high-importance services from being blocked. Simulation results show that compared with the traditional KSP algorithm, the prediction-based dynamic routing intelligent ( P-DRI) algorithm can reduce the service blocking probability by 55.59% , and reduce the average importance of blocking services by 44.77% at the cost of about 6.17% of the calculation delay.

Others

How time latency in navigation Apps affects traffic

Yang Jingjing, Guo Yuchun, Zhao Yongxiang, Chen Yishuai

中国邮电高校学报(英文), 2022, 29 (6). doi： 10.19682/j.cnki.1005-8885.2022.1009

摘要 ( 239 ) PDF (2156 KB)( 24 )

参考文献 | 相关文章 | 计量指标

With the recent increase in routing App usage, it is urgent to analyze the impact of inevitable information delay in routing Apps on traffic. To analyze the negative impact of such delayed routing Apps on traffic, this paper investigates the impact of ideal and delayed routing Apps on traffic. Firstly, the effect of ideal routing Apps has been explored with theoretical analysis and derivation based on a macroscopic network model. Then, an extended network model is built to characterize the delayed routing. The impact of such delayed routing Apps is investigated by simulation experiments while considering routing usage proportion and amount of delay. The results demonstrate that ideal navigation does improve traffic efficiency, but in some cases, delayed navigation is even worse than no navigation.

Low-complexity detection based on maximum cumulative measure and Laurent decomposition for multi-h CPM signal

Zheng Zhidong, Yu Yanming, Liu Ning, Yuan Honggang, Lou Ning

中国邮电高校学报(英文), 2022, 29 (6). doi： 10.19682/j.cnki.1005-8885.2022.1003

摘要 ( 204 ) PDF (501 KB)( 44 )

参考文献 | 相关文章 | 计量指标

In order to solve the problem of high computational complexity in demodulation for multi-h continuous phase modulation (CPM) signal, a maximum cumulative measure combing with the Laurent decomposition (MCM-LD) scheme is proposed to reduce the number of the grid states and the required number of matched filters, which degrades the demodulation complexity at the receiver. The advanced range telemetry (ARTM) Tier 2 CPM signal is adopted to evaluate the performance in simulation. The results show that, compared with the traditional maximum likelihood sequence detection (MLSD), MCM-LD can respectively reduce the numbers of grid states and matched filters from 256 to 32 and 128 to 48 with negligible performance loss, which effectively degrades the computational complexity for multi-h CPM signal.

Face anti-spoofing based on multi-modal and multi-scale features fusion

Kong Chao, Ou Weihua, Gong Xiaofeng, Li Weian, Han Jie, Yao Yi, Xiong Jiahao

中国邮电高校学报(英文), 2022, 29 (6). doi： 10.19682/j.cnki.1005-8885.2022.1004

摘要 ( 333 ) PDF (2578 KB)( 43 )

参考文献 | 相关文章 | 计量指标

Face anti-spoofing is used to assist face recognition system to judge whether the detected face is real face or fake face. In the traditional face anti-spoofing methods, features extracted by hand are used to describe the difference between living face and fraudulent face. But these handmade features do not apply to different variations in an unconstrained environment. The convolutional neural network (CNN) for face deceptions achieves considerable results. However, most existing neural network-based methods simply use neural networks to extract single-scale features from single-modal data, while ignoring multi-scale and multi-modal information. To address this problem, a novel face anti-spoofing method based on multi-modal and multi-scale features fusion ( MMFF) is proposed. Specifically, first residual network ( Resnet )-34 is adopted to extract features of different scales from each modality, then these features of different scales are fused by feature pyramid network (FPN), finally squeeze-and-excitation fusion ( SEF) module and self-attention network ( SAN) are combined to fuse features from different modalities for classification. Experiments on the CASIA-SURF dataset show that the new method based on MMFF achieves better performance compared with most existing methods.

Research on ECG classification based on transfer learning

Jiang Fan, Chen Jiajun, Gao Youjun, Sun Changyin

中国邮电高校学报(英文), 2022, 29 (6). doi： 10.19682/j.cnki.1005-8885.2022.1008

摘要 ( 252 ) PDF (2273 KB)( 36 )

参考文献 | 相关文章 | 计量指标

Concerning current deep learning-based electrocardiograph ( ECG) classification methods, there exists domain discrepancy between the data distributions of the training set and the test set in the inter-patient paradigm. To reduce the negative effect of domain discrepancy on the classification accuracy of ECG signals, this paper incorporates transfer learning into the ECG classification, which aims at applying the knowledge learned from the raining set to the test set. Specifically, this paper first develops a deep domain adaptation network ( DAN) for ECG classification based on the convolutional neural network ( CNN). Then, the network is pre-trained with training set data obtained from the famous Massachusetts Institute of Technology-Beth Israel Hospital (MIT-BIH) ECG arrhythmia database. On this basis, by minimizing the multi-kernel maximum mean discrepancy (MK-MMD) between the data distributions of the training set and the test set, the pre-trained network is adjusted to learn transferable feature representations. Finally, with the low-density separation of unlabeled target data, the feature representations are more transferable. The extensive experimental results show that the proposed domain adaptation method has reached a 7. 58% improvement in overall classification accuracy on the test set, and achieves competitive performance with other state-of-the-arts.