宽带用户行为分析与探究

宽带网络发展到一定的阶段,宽带用户的行为将对网络产生一定的影响,为保障网络健康高效的发展,我们不仅要关心网络运行质量,还要关心用户行为,关注网络流量。本文从用户接入控制,非法接入监控,VOIP监控,P2P监控四个方面进行阐述。     

P2P文件共享网络的不良信息监测

P2P技术作为一种新技术在为信息传递提供更宽广的舞台的同时,也给不良信息传播提供了新途径。近年来,P2P网络上的淫秽、色情和反动等有害信息传播盛行,几乎不受任何限制,这使得监管机构遭遇到前所未有的难题。由于技术手段的落后,如何时P2P网络上不良信息实施监控亟待解决和完善。Gnutella网络是无结构化P2P网络的代表,其同时在线的用户数目已超过100万。本文根据Gnutella网络结构的特点,对Gnutella网络上所共享的淫秽色情信息进行了实时监测,并对Gnutella用户所感兴趣的内容进行了相应的分析,为P2P文件共享网络的不良信息监控进行了初步的尝试。     

基于兴趣社会网络的电子商务平台设计

本文将P2P技术和社会网络分析方法相结合,提出了分布式、高效的基于兴趣社会网络的电子商务平台。给出了电子商务社会网络的整个架构,同时,提出了构成底层P2P电子商务网络的网络拓扑构建方法,提出了根据用户存储的资料和用户行为进行P2P信息搜索机制,并在此基础上提出了根据用户兴趣计算用户之间相似度以形成具有相同兴趣的自组织社会网络。本文结论为基于P2P架构的电子商务实际运营提供了参考。     

基于用户行为分析的PPTV点播系统测量研究

利用基于协议分析和逆向工程的主动测量方法对PPTV点播系统进行了研究,通过对PPTV点播协议进行分析,获悉该协议的通信格式和语义信息,总结出了PPTV点播系统的工作原理,在此基础上设计并实现了基于分布式网络爬虫的PPTV点播系统主动测量平台,并对该平台获取的用户数据进行统计分析,获得了PPTV点播系统部分用户行为特征.研究结果对P2P点播系统的监控及优化提供了研究方法.     

P2P网络视频监控技术机制解析

当前,民用市场中中小企业和家庭视频监控规模呈爆发式增长,针对上千万摄像机的监控需求,产业界普遍采用点对点（peer to peer,P2P）网络视频监控技术,综合阐述了P2P网络视频监控技术原理,分析了主流的P2P 网络视频监控所采用的流媒体控制技术、传输技术以及私网穿越技术,最后提出了大规模部署的P2P视频监控解决方案建议。     

一种基于聚类算法的P2P流媒体服务平台可视化监控系统的设计与实现

基于P2P技术的流媒体服务技术已得到广泛应用．然而P2P流媒体服务系统中拓扑结构的动态性与用户行为的自主性使得流媒体服务质量面临挑战．本文以南京大学研制的交互式P2P流媒体服务平台NJUVoD为背景,基于协同过滤机制及聚类算法,设计了一个面向P2P流媒体服务平台的可视化Web监控系统．实现了对用户行为、拓扑结构、系统状态的在线分析,为了解系统运行状况、优化系统性能提供了决策依据．     

打造完善的操作行为审计解决方案

如何感知用户的网络行为,实现对网络用户行为的审计,并根据审计结果对网络用户行为进行控制,一直是网络管理者关注的问题。远程接入的企业员工做过什么操作;企业的外包服务商远程接入企业平台,做过什么操作;谁更改过服务器的配置,修改过哪些服务器的配置;     

用户端网络远程监控服务促电信网运维转型

电信运营商开展的用户端网络远程监控服务是为用户的广域网提供“一点接入,全网监控”的服务,在用户的网络发生故障的第一时间主动告警,协调解决故障,并定期给用户提供分析报告。该服务在     

基于以太网的宽带接入解决方案

小区宽带数据接入网是广电网发展数据业务时规划的重点。随着数据网络建设的发展 ,接入用户也在不断增加 ,网络结构日趋庞大 ,作为网络运营者 ,如何保证社区网络的安全性 ,如何提高网络访问的响应速度 ,如何在运营网络时保证最高的性能价格比 ,如何为用户提供更好的多功能服务 ,是广电面临的首要问题。1　规划及设计原则(1)用户比较密集的小区、办公楼及其他集团用户 ,采用以太网方式接入。(2 )接入层负责提供各种类型用户的接入 ,较高的端口密集 ,能对用户节点实现一定的流量控制及其他策略管理功能。(3)接入节点主要对小区用户提供接入服…     

基于PTN的动环监控系统更适应IP网络

采用新的PTN传输组网接入技术,不仅网络架构更为简单清晰,维护方便快捷高效,并且可以有力地保障各类高带宽监控需求,同时使接入其他行业用户监控业务成为可能。     

基于电信大数据的5G网络海量用户复访行为预测模型

5G网络中的用户会产生大量的访问数据，导致用户复访行为难以精准预测，因此提出基于电信大数据的5G网络海量用户复访行为预测模型。从电信大数据中提取用户上网历史行为特征数据，构建数据集。引入多阶加权马尔可夫链模型，通过计算各阶自相关系数，得到模型权重值，计算模型的统计量。经过分析后得到各阶步长的马尔可夫氏链一步转移概率矩阵，从而实现对5G网络海量用户复访行为的精准预测。实验结果表明，该模型拥有最低的均值误差和标准差，以及最高的精度、查全率、查准率、F1指标，可证明该方法在预测用户复访行为方面有着非常明显的优势。     

对等环境中访问控制的方法

所有连接到网络中的用户都可以共享网络资源,在这种情况下,对于各种网络应用的攻击也是随时存在的,所以需要更为复杂的安全服务。本文提出一个使用SPKI证书在对等环境中进行访问控制的方案,这个方案是通过给网络中想要共享资源的用户提供证书的方法来进行访问控制的,根据分配给用户的访问权限来限制信息提供者的资源的使用。     

Dynamic bandwidth allocation for quality-of-service over Ethernet PONs

Ethernet-based passive optical network (EPON) technology is being considered as a promising solution for next-generation broadband access networks due to the convergence of low-cost Ethernet equipment and low-cost fiber infrastructures. A major feature for this new architecture is the use of a shared transmission media between all users; hence, medium access control arbitration mechanisms are essential for the successful implementation of EPON: i.e., to ensure a contention-free transmission and provide end users with equal access to the shared media. We propose to use the multipoint control protocol defined by the IEEE 802.3ah task force to arbitrate the transmission of different users, and we present different dynamic bandwidth allocation (DBA) algorithms to allocate bandwidths effectively and fairly between end users. These DBA algorithms are also augmented to support differentiated services, a crucial requirement for a converged broadband access network with heterogeneous traffic. We show that queueing delays under strict bandwidth allocation algorithms result in an unexpected behavior for certain traffic classes, and we suggest the use of DBA with appropriate local queue management to alleviate this inappropriate behavior. We conduct detailed simulation experiments to study the performance and validate the effectiveness of the proposed protocols.     

Resource planning and incentive engineering for a congested wireless access point: an integrated multiple time scale control mechanism

Wireless networks are playing an increasingly important role for global communications. Many resource allocation mechanisms have been proposed to efficiently utilize the limited radio resources in wireless networks to support a large number of mobile users with a diversity of applications. Among them, pricing frameworks that provide incentives to users to maximize their individual utility while optimizing allocation of network resources have attracted a lot of attention recently. Nevertheless, most of these pricing schemes require dynamic charging rates and may be too complex for wide acceptance by users, as most users would prefer relatively simple charging schemes. Moreover, use of a pricing framework to facilitate resource planning and future expansion at the service provider’s side has not yet been widely considered. In this paper, we propose Integrated Multiple Time Scale Control (IMTSC), a novel incentive engineering mechanism to facilitate resource allocation and network planning. Over different time scales, IMTSC combines the functions of network capacity planning, admission control for resource allocation, and tracking of users’ instantaneous traffic demands. The proposed mechanism is applied for access control at a congested access point in a wireless network. By decomposing the original problem into distributed optimization problems that are solved locally by the service provider through adjusting charging rate and remotely by individual users by appropriately changing her service requests, we show that maximization of user’s utility and increase of network efficiency can be simultaneously achieved. Results from extensive simulations demonstrate the effectiveness of the proposed IMTSC mechanism.     

Hadoop cloud platform user dynamic access control model

In order to solve the problem that Hadoop cloud platform could not dynamically control user access request,a Hadoop cloud dynamic access control model based on user behavior assessment (DACUBA) was proposed.The model first collected the user instruction sequence in real time and the user behavior contour was obtained by parallel command sequence learning (PCSL).Then the global K model was established by using the forward profile,the subsequent sequence was classified and the classification results were evaluated.The evaluation results were combined with the improved Hadoop access control mechanism to make the cloud platform users’ access rights change dynamically with their own behaviors.Experimental results demonstrate that the model algorithm is effective and the dynamic access control mechanism is feasible.     

基于非合作博弈的访问控制安全机制

访问控制技术是保护计算机系统和网络中的敏感信息和关键资源的一种重要的信息安全技术。但是传统的访问控制方法只能被动地对用户的访问请求进行响应,无法完全适应如今不断动态变化的网络环境。为了提高访问控制中系统应对恶意攻击和威胁的能力,基于博弈论中的非合作博弈方法,设计了一套适用于访问控制模型的安全机制,当访问主体提出访问请求后,对访问主体与被访问客体进行非合作博弈,得到纳什均衡,最终通过纳什均衡与门限的比较决定是否对访问主体进行授权。实验证明了该安全机制的有效性。     

基于2-D稳定条件的混合流网络参数设置

由于网络容量的限制,任何一个网络都不可能避免拥塞问题。传统的RED算法只考虑了少量TCP用户,没有涉及UDP用户的情况,同时随着网络应用的多样化,如越来越多的UDP用户接入网络,传统的RED机制无法控制它们,保证不了TCP用户的服务质量(QoS)。针对TCP/UDP混合流多用户的情况,本文提出TCP/UDP混合流的区分控制,这里TCP和UDP流使用不同的带宽,及TCP/UDP混合流单瓶颈网络的2-D稳定条件。基于该稳定条件可以选择一个合适的RED控制参数Pmax,获得满意的网络拥塞控制性能。本文建立了一个TCP和UDP流单瓶颈网络的线性时滞系统模型,利用2-D拉普拉斯-Z变换,推导出基于稳定条件的混合流网络参数配置。利用NS2仿真验证所提出的混合流网络参数配置能够有效获得关于路由器队列长度和TCP窗口的稳定性。     

A Human-in-the-Loop Architecture for Mobile Network: From the View of Large Scale Mobile Data Traffic

Unlike other radio signal services, 5G is anticipated to play a huge role in offering services to heterogeneous networks, technologies, and devices operating in different geographic regions to fulfill the high expectation of users with relatively low energy consumption, which implies the necessity for moving from a system-centric design to a more user- or even human- and data- centric design paradigm “to keep the human in the loop” in future network. It drives us to design a system with capacity to allocate network resource dynamically according to feedback from users. This paper presents a Human-In-The-Loop architecture for mobile network that discovers users’ needs on network resource by understanding data traffic usage behavior of users. Based on real data traffic of mobile network, we analyze data traffic patterns of heavy and normal users from the view of online browsing behavior and urban functional area to explain how and why the data traffic is consumed. Then we propose a Latent Dirichlet Allocation model based solution to correlate data traffic, user behavior, and urban ecology to gain deep insights into spatio-temporal dynamic of data traffic usage behavior for different groups of users. Drawing upon results from a comprehensive study of users in a metropolitan city in China, we achieve a broad understanding about the difference of data traffic usage patterns of heavy and normal user: (1) besides the amount of generated data traffic, two groups of users can be easily distinguished by usage behavior of limited number of applications at midnight, (2) the functions of locations have huge impact on data usage patterns of users, which implies that urban ecology will shape users’ online behavior. The results of this work can potentially be exploited to help to allocate network resource, improve Quality of Experience according to users’ needs, and even design the future network.     

Dynamiques de réseau et externalités de demande une approche en termes ďéquilibre bayésien parfait *

This paper analyses the dynamics of a network size when the value assigned to this network by the potential users depends on the number of subscribers. Each potential subscriber’s network value is an individual parameter, known only to that subscriber. This situation is typically a coordination problem in incomplete information. The observation of each other move, when this situation is repeated, can reduce the initial uncertainty about player’s types resulting in a network dynamics. We show that the repetition makes the access to the network easier, when the users are interested by connexion, and more difficult, when the users are not interested by entering the network.     

CAC-RD: an UMTS call admission control

The third wireless network generation (3G) aims to provide fast Internet access with quality of service (QoS) guarantees, especially to multimedia applications. UMTS (Universal Mobile Telecommunication System) is a kind of 3G networks. To provide QoS, the network must use an efficient admission control mechanism. This mechanism needs to prioritize network access to critical classes of applications. This work proposes an UMTS admission control mechanism, called CAC-RD (Call Admission Control—based on Reservation and Diagnosis). It is based on network diagnosis and on channel reservation for handovers. These techniques are associated with new calls blocking when the network reaches utilization thresholds. CAC-RD is a tool that prioritises handovers and conversational applications. The main CAC-RD goals are the handovers blocking reduction and the acceptable performance levels guarantee. Simulation results show that the CAC-RD channel reservation and the diagnosis techniques associated with the intrinsic network signal power control effectively reduces blockings while guarantying performance levels. Due to computational resource limits, simulations cannot answer related to admission control in big networks with thousands of users. This work presents a method to extrapolate scientific questions like CAC’s behavior with thousands of users and many antennas. An artificial neural network approach for CAC-RD in UMTS 3G networks is presented as an extension of the work.