SDON中路径建立和资源分配研究北大核心

从软件定义光网络(SDON)的路径建立和资源分配问题着手,首先描述了SDON控制平面面向业务的特点并对资源抽象过程进行了描述,接着研究了Open Flow协议向光域扩展的策略,最后分析了SDON中路径建立和资源分配过程及对应的关键技术,着重对不同技术域内的路径建立和资源分配策略进行了研究。     

基于SDON的智能专线业务探讨

以目前广受关注的软件定义网络为背景,首先介绍其在光传输网络领域的应用——软件定义光网络,然后探讨在软件定义光网络的基础上如何部署智能专线业务,实现网络能力开放,达到快速开通业务、灵活调整带宽以及简化运营维护过程的目的,使专线业务的智能化成为可能.     

基于图神经网络的SDON性能预测模型

网络性能预测是实现软件定义光网络(SDON)高效网络管理的关键,但目前亟需一种能够以较低成本准确预测关键指标的网络性能预测模型。提出一种基于图神经网络的SDON性能预测模型,该模型将Bi GRU和Self-Attention机制相结合,能够学习网络拓扑、路由和流量矩阵之间的复杂关系,从而准确地估计网络中源/目的地的分组延迟、抖动以及丢包率,并且可以应用于训练中未遇到的网络。实验结果表明,在不同流量模型测试中,所提模型相较于基线模型的平均绝对百分比误差(MAPE)性能有明显提升。     

在SDON中面向成本优化的虚拟光网络映射方法

结合软件定义光网络和光网络资源虚拟化的技术背景,对虚拟光网络成本优化问题进行了深入研究。在软件定义光网络中采用软件定义光网络基础服务架构和网络资源抽象的虚拟光网络服务架构,通过引入虚拟光网络的映射模型,提出了虚拟光网络映射策略和虚拟光网络协同映射规划方法,解决了面向成本优化的虚拟光网络映射方法,可为未来光网络虚拟化技术的应用提供技术支撑。     

软件定义光网络(SDON)发展前瞻

为了适应宽带业务与应用的持续发展,光网络需要变得更加灵活可控。软件定义光网络(SDON)聚焦于将软件定义技术融入光网络的综合解决方案,其关键技术涉及软件定义光传输、交换和联网等。通过分析SDON的技术特征,探讨了SDON引入光网络的必要性和可行性,设计了SDON的典型应用案例,并可以预见SDON在支撑新一代光网络向智能高效、灵活调度、虚拟重构等方向的发展进程中将发挥重要作用。     

基于LSTM流量预测的路由规划和切换

随着近年网络技术的不断发展和演进,用户对网络资源的需求体量越来越大,内容越来越复杂,传统的网络架构很难满足当前灵活的网络需求。软件定义网络的革命性技术思想赋予了网络可编程性和可演进性,实现网络资源更灵活的管理。为了提高网络资源利用率,提出了基于LSTM模型的流量预测,为路由规划提供理论依据。通过提前预测网络流量,在流量激增之前进行防御措施保证网络的安全性和稳定性。实现基于流量预测的路由规划、无损路径切换等功能。     

基于贝叶斯算法的XG-EPON业务感知机制

提出了一种基于贝叶斯算法的XG-PON业务感知机制,该机制采用贝叶斯算法对业务进行分类,以实现高效的业务感知。在此基础上通过OLT与ONU的交互,优化XG-PON的资源分配。对基于贝叶斯算法的XG-PON业务感知机制进行仿真,仿真结果表明该机制能有效降低丢包率并缩短时延,提高了系统对多业务的支持能力。     

拟态SDON传输系统中冗余异构多核架构的实现

现有增强光传输网络安全性的方法多是基于网络层面的,无法保障光传输设备的安全性.提出在光传输设备的管理控制系统中引入拟态防御技术,设计了设备的冗余异构多核架构.基于系统相似度构建攻击概率模型,并进行仿真验证.结果 表明:所提方法明显降低了系统被攻击成功的概率,大大提高了拟态软件定义光网络传输系统的安全性能.     

智能光网络发展历程的回顾和展望:从ASON到PCE,再到SDON

回顾了智能光网络的发展历程,并根据其多域异构化的发展趋势,分析和比较了现有典型光网络架构的特点和性能,进而综合其各自优点,提出了基于PCE和域间连接控制的多域异构光网络管控架构PIONEER以及面向未来可灵活配置、扩展和升级的软件定义光网络(SDON)架构。     

PTN流量预测及控制策略研究

PTN(分组传送网)是基于IP(互联网协议)技术的新一代分组光传送网技术,文章分别对城域PTN的核心层、汇聚层和接入层的流量预测建立了相应的数学模型,基于模型进行了流量预测和带宽测算,提出了PTN流量控制的相关策略,对于指导PTN的建设、运维期间的业务开通和业务配置工作有一定的参考价值。     

A wavelength-buffering scheme for dynamic traffic in optical wavelength-division multiplexing networks

In optical wavelength-division multiplexing (WDM) networks, traffic can be very “bursty” at a fine time scale, even though it may seem to be smooth at coarser scales (e.g., Poisson or Poisson-related traffic). This paper analyzes the instantaneous characterization of Poisson traffic at a fine time scale. The analysis shows that the irregular oscillation of the instantaneous traffic load and the occurrence of blockings in a light-loaded network are highly correlated. Specifically, most blockings occur concentratively at the peaks of the instantaneous load. In some other time, network resources may not be sufficiently utilized. To make better utilization of network resources, a novel wavelength-buffering (WB) scheme is proposed for the first time in this paper. By reserving a portion of resources in a “wavelength buffer” under light loading and releasing them when the load goes up, a number of blockings brought by the oscillation of the traffic load can be avoided. Simulation results show that compared with other schemes such as adaptive routing, wavelength conversion (WC), and rerouting, the novel wavelength-buffering scheme achieves significantly better performance with respect to the network utilization and overall blocking probability.

基于自回归模型与神经网络模型的车流量预测对比

车流量建模是车联网(vehicular Ad Hoc network,VANET)路由、多媒体接入协议、无线算法设计的基础.准确的车流量模型将对智能交通系统(intelligent transportation system,ITS)实时调度和车联网的信息安全起到十分重要的作用.基于上海市的交通流量数据,利用自回归(auto regressive,AR)模型与神经(back-propagation,BP)网络模型对车流量实测数据进行了仿真对比,给出了相应的预测结果.研究发现,两个模型均能有效地对数据进行跟踪与预测,但对不同时段数据预测的准确性有所不同.研究结果将为未来智能交通应用、车联网的理论研究等提供有力依据.     

Optimized traffic flow prediction based on cluster formation and reinforcement learning

In recent days, the traffic flow information is collected using the global positioning system through the Internet, which is yet to become ubiquitous. A novel technique is proposed for the intelligent transportation system, which leads to reduce the traffic congestion that will become an unavoidable phenomenon in the near future. This system uses a magnetic sensor to identify the type of the vehicle and the exact vehicle count in the traffic environment based on variation in the magnetic flux. This information is transmitted to the cloud server with the help of cluster by utilizing the nearby proximity services. An intelligent agent that uses reinforcement learning is implemented in the cloud server to learn the real-time traffic flow from multiple sources for the prediction of a valid and optimized route suggestion for the registered users. This work is implemented, and implementation results show that the proposed work achieves an accuracy of 98.36%. Hence, this intelligence method for VANETs will certainly account for improved traffic prediction to the vehicle transportation. It can reduce the vehicles waiting time in traffic and that would minimize the fuel consumption. It will make an eco-friendly environment of reduced carbon dioxide emissions in urban cities.     

ABR业务流量控制的一种智能机制

本文提出了一种ＡＢＲ业务模糊流量控制机制,仿真结果表明它有效地避免了网络阻塞,在网络的吞吐量上要高于强比例速率控制算法（ＥＰＲＣＡ）。该机制与连接业务的特性无关,且不需要改变有关ＡＢＲ闭环反馈的流量控制结构,这为算法的实际应用提供了前提。     

IP骨干网流量的自动化预测

高效、可靠的网络流量预测是网络规划、扩容建设的基础。互联网流量目前缺乏完备的理论模型,行业内大多根据工程实践特点,设计简化可操作的预测模型以满足IP网络规划需求。首先根据中国电信自身IP骨干网流量预测工作的需求及特点,使用时间序列分析的多因子回归模型和函数自适应模型对IP骨干网流量进行分析和预测,基于大量现网实际数据的仿真运算,对比两种模型的特点、优劣和适用场景,提出了一种预测模型选择和参数优化的原则和方法。在此基础上,构建了可以满足百千量级时间序列要求的自动化流量预测系统,极大简化并提升了流量预测工作的效率。最后,展望了未来IP流量预测工作的延展方向和关注重点。     

一种基于频繁序列匹配的交通状态趋势预测方法

海量的交通流数据中一定隐藏着某些潜在的交通状态演变规律,然而少有研究能以实验的方法支持这种观点。本文提出一种基于频繁序列匹配的交通状态趋势预测方法：首先,介绍交通状态序列模型及序列切分处理,其次,提出投影压缩序列相似度计算方法及序列匹配算法,第三,讨论基于频繁序列匹配的交通序列预测算法。通过真实采集数据验证本文提出的方法可行有效。     

Attention-based spatio-temporal graph convolutional network considering external factors for multi-step traffic flow prediction

Traffic flow prediction is an important part of the intelligent transportation system. Accurate multi-step traffic flow prediction plays an important role in improving the operational efficiency of the traffic network. Since traffic flow data has complex spatio-temporal correlation and non-linearity, existing prediction methods are mainly accomplished through a combination of a Graph Convolutional Network (GCN) and a recurrent neural network. The combination strategy has an excellent performance in traffic prediction tasks. However, multi-step prediction error accumulates with the predicted step size. Some scholars use multiple sampling sequences to achieve more accurate prediction results. But it requires high hardware conditions and multiplied training time. Considering the spatiotemporal correlation of traffic flow and influence of external factors, we propose an Attention Based Spatio-Temporal Graph Convolutional Network considering External Factors (ABSTGCN-EF) for multi-step traffic flow prediction. This model models the traffic flow as diffusion on a digraph and extracts the spatial characteristics of traffic flow through GCN. We add meaningful time-slots attention to the encoder-decoder to form an Attention Encoder Network (AEN) to handle temporal correlation. The attention vector is used as a competitive choice to draw the correlation between predicted states and historical states. We considered the impact of three external factors (daytime, weekdays, and traffic accident markers) on the traffic flow prediction tasks. Experiments on two public data sets show that it makes sense to consider external factors. The prediction performance of our ABSTGCN-EF model achieves 7.2%–8.7% higher than the state-of-the-art baselines.     

结合多重分形的网络流量非线性预测

通过分析树型多重分形结构的相关性发现，多重分形可以把非平稳且具有长相关（LRD）和分形特性的网络流量序列转化为可用短相关（SRD）模型表示的序列组。利用多重分形这种将时间序列分解为多层的能力，提出了一种结合多重分形的FIR神经网络流量预测模型（MF-FIR，multifractal FIR network）。MF-FIR合理地利用了流量序列的LRD信息，具有很好的多步预测性能，可以满足通信系统在线预测的要求。     

Identifying online traffic based on property of TCP flow

Classification of network traffic using port-based or payload-based analysis is becoming increasingly difficult when many applications use dynamic port numbers, masquerading techniques, and encryption to avoid detection. In this article, an approach is presented for online traffic classification relying on the observation of the first n packets of a transmission control protocol (TCP) connection. Its key idea is to utilize the properties of the observed first ten packets of a TCP connection and Bayesian network method to build a classifier. This classifier can classify TCP flows dynamically as packets pass through it by deciding whether a TCP flow belongs to a given application. The experimental results show that the proposed approach performs well in online Internet traffic classification and that it is superior to naive Bayesian method.