基于FARIMA-GARCH模型的网络业务预测算法

网络流量的波动性与自相似特性为其精确预测提出了挑战。为此,提出了一种基于FARIMA-GARCH模型的预测算法。该算法首先利用分段双向CUSUM检测算法对流量序列的均值进行有效检测,并在此基础上将序列零均值化;然后采用限定搜索法对分数差分阶数进行精确估计;在获得模型参数后,使用FARIMA-GARCH模型对网络流量进行预测。仿真实验表明,限定搜索法能够获得比传统算法更高的估计精度。随后采用真实网络流量验证了预测算法的性能,在保持与FARIMA预测算法等价的时间复杂度下,其均方根和相对均方根误差与RBF神经网络预测算法相当,而高于FARIMA预测算法。同时,预测算法对突发流量的跟踪和预测性能明显优于对比算法,且有更好的区间估计性能。     

改进的基于时变FARIMA模型和小波变换的网络流量预测算法

网络的流量特性是反映网络实时状态的一个重要特征,对于网络流量的分析、预测一直是该领域的研究热点。传统的基于时间序列模型的方法在计算效率和多尺度分析能力方面存在一定的局限性。本文提出了一种改进的基于小波变换和时变FARIMA模型的流量预测方法,利用小波变换的多尺度分析特性将原有的流量数据进行分解,在使用时变FARIMA模型进行预测,可大大提高算法的执行效率和预测的准确性。最后,本文选取了Bellcore提供的真实的网络流量进行了仿真实验,验证了提出的预测方法的准确性和有效性。     

改进的基于小波变换和FARIMA模型的网络流量预测算法

提出了一种改进的基于小波变换和FARIMA模型的网络流量预测算法,先对经过预处理的流量进行小波分解,再进行Mallat算法单支重构,接着用FARIMA模型分别对重构后的单支进行预测,最后合成流量。该算法较之传统的首先用FARIMA模型对小波系数进行预测再进行小波重构的算法减小了预测误差。仿真实验也验证了改进算法的预测准确性。     

网络流量预测中的时间序列模型比较研究

网络流量预测在新一代网络协议设计、网络管理与诊断、设计高性能路由器等方面都具有重要意义.目前通常采用ARMA和FARIMA时序模型对网络流量序列进行拟合与预测,但没有对时间尺度的大小与模型选择的关系进行研究.本文对实际网络流量在不同时间尺度(毫秒、秒、分)下进行了流量预测建模并对预测性能进行比较,分析表明使用时序模型进行流量预测时,大时间尺度(分)流量预测较小时间尺度(毫秒、秒)具有更小的预测误差.并且,对于小时间尺度上的自相似流量序列,自相似模型FARIMA并没有较其他时序模型有更好的预测性能.     

基于FARIMA过程的网络业务预报与应用

在高速网络控制与带宽分配研究中,网络业务量预报是一个重要的问题。本文首先介绍自回归分数整合滑动平均（FARIMA,Farctal Autoregressive Integrated Moving Average)过程的概念及其具体形式,并给出了FARIMA过程的预报方法,然后基于FARIMA过程的最小均方误差预报方法,提出了一个具有补偿功能的网络自相似业务的预报方法,最后给出这种预报方式在网络控制研究中的应用。     

小波系数感知的网络流量预测机制

对网络流量的精确预测,可以准确把握网络运行趋势,及时防范网络故障。针对长期网络流量预测准确度低,收敛速度慢的问题,提出一种小波系数感知的网络流量预测(WCNTP)机制。借助重标极差(R/S)序列分析法初步评估网络流量在大时间尺度上的统计特性;利用离散小波变换将非平稳的网络流量分解为多个相对平稳的流量序列;利用分数自回归求和滑动(FARIMA)模型对网络流量进行预测。结果表明,所提机制在长期网络流量预测过程中,具有较高的准确度且收敛速度快,能够精确评估网络性能,在保证网络平稳运行的同时,提高网络服务质量。     

Network traffic prediction method based on wavelet transform and multiple models fusion

Accurate prediction of network traffic is an important premise in network management and congestion control. In order to improve the prediction accuracy of network traffic, a prediction method based on wavelet transform and multiple models fusion is presented. Mallat wavelet transform algorithm is used to decompose and reconstruct the network traffic time series. The approximate and detailed components of the original network traffic can be obtained. The characteristics of approximate components and detail components are analyzed by Hurst exponent. Then, according to the different characteristics of the components, autoregressive integrated moving average model (ARIMA) is chosen as the prediction model for the approximate component. Least squares support vector machine (LSSVM) is used to predict detail component. Meanwhile, an improved particle swarm optimization (PSO) algorithm is proposed to optimize the parameters of the LSSVM model. Gauss‐Markov estimation algorithm is adapted to fuse the predicted values of multiple prediction models. The variance of fusion prediction error is smaller than that of single prediction model, and the prediction accuracy is improved. Two actual datasets of network traffic are studied. Compared with other state‐of‐the‐art models, the case study results indicate that the proposed prediction method has a better prediction effect.     

MIMO ARIMA models for handoff resource reservation in multimedia wireless networks

This paper presents a new approach to prediction of resource demand for future handoff calls in multimedia wireless IP networks. Our approach is based on application of multi‐input‐multi‐output (MIMO) multiplicative autoregressive‐integrated‐moving average (ARIMA) (p,d,q)x(P,D,Q)_S models fitted to the traffic data measured in the considered cell itself and on the new call admission control (CAC) algorithm that simultaneously maximizes the system throughput while keeping the handoff call dropping probability (CDP) below the targeted value. The main advantages of the proposed approach are the following: first, the proposed multi‐variable prediction method gives on average better predictions (i.e. narrower prediction confidence interval) for realistic traffic situations, which results in lower new call blocking probability (CBP) at the targeted handoff CDP and second, the model is simple to implement since it does not require communication among the adjacent cells. Simulation results show the superiority of the proposed MIMO prediction approach combined with the proposed call admission control algorithm for some typical nonstationary situations in comparison with univariate models. Copyright © 2004 John Wiley & Sons, Ltd.     

基于混沌理论与改进回声状态网络的网络流量多步预测

网络流量预测是网络管理及网络拥塞控制的重要问题,针对该问题提出一种基于混沌理论与改进回声状态网络的网络流量预测方法。首先利用0-1混沌测试法与最大Lyapunov指数法对不同时间尺度下的网络流量样本数据进行分析,确定网络流量在不同时间尺度下都具有混沌特性。将相空间重构技术引入网络流量预测,通过C-C方法确定延迟时间,G-P算法确定嵌入维数。对网络流量时间序列进行相空间重构之后,利用一种改进的回声状态网络进行网络流量的多步预测。提出一种改进的和声搜索优化算法对回声状态网络的相关参数进行优化以提高预测精度。利用网络流量的公共数据集以及实际数据进行了仿真,结果表明,提出的预测方法具有更高的预测精度以及更小的预测误差。     

互联网流量与两个相关模型

由于互联网具有多构性、突发连续性和自相似性等特征,使得用传统模型进行的排队分析、性能估计与实际网络有较大差距。本文介绍了两种互联网流量模型:AR IMA模型,FAR IMA模型,讨论了它们的适用范围、模型数学定义,描述了参数定阶推导等相关问题。     

ATM网络异种复用业务信元丢失率快速估计的新算法

ＡＴＭ网络信元丢失率（ＣＬＲ）的估计是呼叫接入和流量控制的关键技术,基于信元丢失机制的分析,本文了一种改进的简单业务模型,并进而得到一种新的算法,能够对异种混合业务复用的ＡＴＭ网络进行快速的ＣＬＲ估计,该算法处理速度快,能够做到呼叫的实时响应,模型采用国际规范的标准参数构造,可以直接应用于实际操作,仿真结果表明,算法的精度,运算复杂度和算法鲁棒性都比较理想,具有较高的实用价值。     

自相似模型下的交换式网络仿真

传统模式下的网络仿真,报文到达均服从的是一种具有短相关特性的泊松分布。而经过大量业务流量监测表明,网络流量实际呈现出的确是一种具有长相关特性的自相似分布,这种特性对网络流量建模、性能分析、接纳控制等产生了重要影响。在对自相似特性深入分析的基础上,利用分型布朗运动模型的RMD算法产生自相似序列来模拟网络业务,并对该业务流特性下的交换式以太网进行了仿真实验。结果表明业务量的自相似性对交换式网络的各项性能影响很大,这与传统流量模型形成鲜明对比。     

A combined admission control algorithm with DA protocol for satellite ATM networks

Admission control is an important strategy for Quality of Service （QoS） provisioning in Asynchronous Transfer Mode （ATM） networks. Based on a control-theory model of resources on-Demand Allocation （DA） protocol, the paper studies the effect of the protocol on the statistical characteristics of network traffic, and proposes a combined connection admission control algorithm with the DA protocol to achieve full utilization of link resources in satellite communication systems. The proposed algorithm is based on the cross-layer-design approach. Theoretical analysis and system simulation results show that the proposed algorithm can admit more connections within certain admission thresholds than one that does not take into account the DA protocol. Thus, the proposed algorithm can increase admission ratio of traffic sources for satellite ATM networks and improve satellite link utilization.     

Rate control of elastic connections competing with long-rangedependent network traffic

Long-range dependence is regarded as a fundamental property of network traffic. Using an original approach, we incorporate this property in a traffic control mechanism for elastic connections that can adapt to the instantaneous network load in a differentiated services-type framework. In this scenario, the network makes predictions of bandwidth requirements of the high-priority traffic and returns feedback information to the elastic source. We include a prediction compensation algorithm that compensates for the larger prediction errors for connections with longer roundtrip delay, and analyze the performance of this algorithm. The specific topology involved in traffic control for differentiated services is thus harnessed, together with the long-range dependence, to improve network performance, thereby counteracting the undesirable characteristics of self-similarity. Furthermore, an adaptive version of the rate-based control algorithm is studied, based on the use of real-time estimates of traffic parameters, including the mean, variance, and Hurst parameter     

An adaptive network traffic prediction approach for LDoS attacks detection

Low‐rate denial of service (LDoS) attacks reduce throughput and degrade quality of service (QoS) of network services by sending out attack packets with relatively low average rate. LDoS attack flows are difficult to detect from normal traffic since it has the property of low average rate. The research on network traffic analysis and modeling shows that network traffic measurement data are irregular nonlinear time series. To characterize and analyze network traffic between attack and non‐attack situations, the adaptive normal and abnormal ν‐support vector regression (ν‐SVR) prediction models are constructed on the basis of the reconstructed phase space. In this paper, the dimension of reconstructed phase space for ν‐SVR is optimized by Bayesian information criteria method, and the parameter in the radial basis function is adaptively adjusted by minimizing the within‐class distance and maximizing the between‐class distance in the feature space. The nonthreshold decision function is obtained through calculating the prediction error of adaptive normal and abnormal ν‐SVR prediction models, which is adopted to detect LDoS attacks. Experiments in NS‐2 environment show that the adaptive ν‐SVR prediction model can effectively predict the network traffic measurement time series, and the probability distribution of time series generated by the adaptive ν‐SVR prediction model is quite similar to that of the network traffic measurement data. Experiments also clearly demonstrate the superiority of the proposed approach in LDoS attacks detection.     

An Online Buffer Management Algorithm for QoS‐Sensitive Multimedia Networks

In this letter, we propose a new online buffer management algorithm to simultaneously provide diverse multimedia traffic services and enhance network performance. Our online approach exhibits dynamic adaptability and responsiveness to the current traffic conditions in multimedia networks. This approach can provide high buffer utilization and thereby improve packet loss performance at the time of congestion.     

Characterization of jitter and admission control in multiservice networks

In statistical admission control based on effective bandwidth, the network provides probabilistic guarantees for the packet loss and delay. In integrated multiservice networks like third-generation mobile networks, some traffic classes also need guarantees on the jitter. One approach is to treat jitter like delay. This letter proposes a new criterion to take into account jitter guarantees. Numerical results show that our criterion is effective and adds improvement up to 10% on the bandwidth management.     

Traffic modelling and forecasting using genetic algorithms for next-generation cognitive radio applications

This article presents a genetic-algorithm-based prediction model for forecasting traffic demands of next-generation wireless networks that are expected to be chaotic in nature. The model approximates the best-fit mathematical equation that generates a given time series using a genetic algorithm. It estimates future traffic in wireless networks using the most recent traffic data points collected from the actual network. Such estimations will be beneficial for network operators helping to manage and optimise the limited radio resources efficiently and eventually to facilitate cognitive radio applications. The new model is compared with conventional regressions analysis and exponential smoothing models, and it has been found that the genetic algorithm model successfully recovers the underlying mathematical expression describing chaotic time series in less than 200 generations and the predictions achieved are by far better than those of regression and exponential smoothing models. The model also offers benefits for in cognitive communication systems with their intrinsic learning capabilities and distributed access decisions.     

基于N阶近邻路网的车辆行程时间估计模型

车联网的提出为智能交通的研究提供了新的交通信息收集技术.针对短时交通中车辆的路网行程时间估计问题,提出了基于N阶近邻的隐马尔科夫模型,利用马尔科夫性质来解决道路行程时间的前后关联性问题,同时考虑不同道路的异构性构建了N阶近邻路网模型来模拟路网间的交互影响.针对短时交通中实时数据更新的问题,提出基于道路关联性算法,并结合车联网的采集技术给出了迭代更新模型的方法.实验表明,本文提出的方法在短时交通车辆行程时间预测中精度较高,能够在车辆行进中做出实时预测.     

基于包容性检验和神经网络的网络流量预测模型

网络流量具有高度复杂的非线性特征,采用单一预测模型往往难以达到理想的预测效果,为此,提出一种包容性检验和BP神经网络相融合的网络流量预测模型（ET-BPNN）。首先采用多个单一模型对网络流量进行预测,然后通过包容性检验,根据t统计量检验选择最合适的基本模型,最后采用BP神经网络对基本模型预测结果进行组合得到最终预测结果。实验结果表明,相对于单一模型以及传统组合模型,ET-BPNN更加准确刻画了网络流量变化趋势,各项评价指标均达到更优,为实现网络流量准确预测提供了更为科学的方法。