DHCP的网络多播克隆在机房管理中的应用

介绍了如何在机房的维护管理中使用Ghost并利用DHCP技术实现网络多播克隆的方法和技巧。包括DHCP服务器的安装与配置,工作站启动盘的制作,硬盘镜像文件的生成和多播克隆的实施,以及巧用Ghost命令行参数,实现无人值守和不同配置机器的多播克隆方法。     

应用层组播时延模型及出度优化算法

为了使应用层组播(ALM)的数据包转发时延能够满足组播应用的要求,根据ALM时延特性建立了ALM时延模型,提出并证明了ALM具有最小平均时延和最小最大时延的ALM最小时延定理,同时根据该定理设计了ALM节点主动调节出度的时延优化算法(AOLA)和被动调节出度的时延优化算法(POLA),两个算法在全局优化策略和局部优化策略下的复杂度分别为O(N)和O(logN).仿真实验表明:所提出的ALM时延模型和ALM最小时延定理是正确的,AOLA和POLA时延优化算法能够显著减小ALM时延,其优化效率最高可达58.76%.     

A multicast routing algorithm with multiple trees

1.INTRODUCTION Qualityofservice(QoS)multicastroutinghascon tinuedtobeaveryimportantresearchtopicintheIn ternet.Multicastisacommunicationservicethatal lowssimultaneoustransmissionofthesamemessage fromonesourcetoagroupofdestinationnodes.To carryoutdifferentmulticastsessions,anetworkmust minimizethesessions’resourceconsumption,while meetingtheirrequirements.Efficientallocationof networkresourcestosatisfyQoSrequirementsisthe primarygoalofmulticastrouting.Howevertheinter dependencyandconfli…     

Distributed QoS multicast routing in networks with imprecise state information

1.INTRODUCTION WiththerapiddevelopmentofInternet,mobilenet worksandhigh performancenetworkingtechnology,itbringsoutextensiveinterestsofpeopleinproviding multimediaandreal timeservicessuchasvideoconfer ence,distanceeducation,entertainmentandsoon.Theseservicesrequirecertainquality of service(QoS)fromthenetworks.Thatis,thereshouldbe guaranteesastobandwidth,delay,delay jitter,ratio ofpacket lossandlowcostduringthedatatransmis sion.Meanwhile,multicastserviceshavebeenused invariousmultimediaap…     

一种用于MPRN的多播路由协议

提出了一种采用按需路由发现策略的移动分组无线网多播路由协议.协议采用了基于网格的方案.在协议中不需要维护全局的路由信息,通过按需路由发现过程,动态更新路由信息和建立多播转发组.多播数据包在转发组成员间进行有限泛洪,使协议能够降低带宽的占用及减少路由更新信息的传播.仿真实验说明,协议在带宽较窄、多播组成员数量变化较大的移动分组无线网环境中,具有较好的可靠性和工作效率.     

一种基于蚁群算法的多媒体网络多播路由算法

为了克服蚁群算法（Ant Colony Optimization,ACO）收敛速度慢，易限于局部最小点等缺陷，对ACO进行了改进，在每次循环结束时，保留最优解，自适应地改变挥发度系数，引入遗传算法的交叉算子，提出了一种基于ACO的有时延约束的多播路由算法模型。仿真结果表明，基于改进ACO的多播路由算法模型 可以稳定地获得优于现有启发式算法的解，是一种有效的多播路算法，该算法也适用于并行执行和应用。     

IP网上可靠组播传输协议(RMSA)的研究

提出了一种新的IP网上可靠组播传输协议（RMSA）。RMSA协议设计方案中，首次使用了通过组播发送方驱动、动态为各个接收节点选择重传请求处理节点的方法。这种方法使得RMSA协议具有较强的可扩展性，较好地适应大规模组播应用中组播树的拓扑结构和网络传输状况发生动态变化的情况。仿真实验表明，RMSA具有差错修复时延较小和重复修复数据较少的优点。同时对现有IP网上路由器组播功能的改动量较现有的方案LMS（lightweight multicast service）和PGM（pragmatic general multicast）少，因此更加接近实用。     

A distributed routing and wavelength assignment algorithm for online multicast in all-optical WDM networks

Routing and wavelength assignment for online real-time multicast connection setup is a difficult task due to the dynamic change of availabilities of wavelengths on links and the consideration of wavelength conversion delay in WDM networks. This paper presents a distributed routing and wavelength assignment scheme for the setup of real-time multicast connections. It integrates routing and wavelength assignment as a single process, which greatly reduces the connection setup time. The proposed routing method is based on the Prim’s MST (Minimum Spanning Tree) algorithm and the K-restricted breadth-first search method, which can produce a sub-minimal cost tree under a given delay bound. The wave-length assignment uses the least-conversion and load balancing strategies. Simulation results show that the proposed algorithm is suitable for online multicast connection establishment in WDM networks. Biography: Huang Chuan-he(1963-), male, Associate professor, research direction: computer networks, distributed computing systems, quantum computation.     

Core Placement Algorithm for Multicast Routing with QoS Requirements

Differing from the source-oriented algorithms,the Core-Based Tree(CBT) multicast routing architecture establishes a single shared tree for multiple connections on a multicast group.Which results in higher ration of network resources utilization.In alluding to the problem of Core Placement,we propose a simple method(QOCP) to locate an optimal core node,which can minimize the multicast delay and inter-destination delay variation simultaneously.The simulation results show that our method is very effective,and outperforms the other algorithms studied in this paper.     

基于NDN的高效发布/订阅系统设计与实现

TCP/IP(transmission control protocol/internet protocol)网络下的发布/订阅系统受限于IP与位置的绑定关系,需要复杂的中间件进行发布者与订阅者之间事件的匹配.命名数据网络(named data networking,NDN)以数据命名取代IP的主体地位,由于其具有上下逻辑拓扑一致性、天然支持多播、具有网络层缓存、多路径转发等特点,故NDN在实现发布/订阅系统上更简洁、高效.设计了一个基于NDN的高效发布/订阅通信协议,并且利用NDN的命名机制对传输小块数据的发布/订阅通信协议进行了进一步优化,使得发布过程的网络开销减少一半.在Android平台上实现了分别代表大块数据和小块数据传输的新闻和天气信息发布/订阅系统,功能验证显示系统支持分布式多发布者、多订阅者的发布和订阅活动,而且对用户操作具有一定的容错功能.在ndnSIM上对所设计的2种发布/订阅通信协议进行模拟,实验结果显示优化后的发布/订阅通信协议在系统效率、传输时延2个性能指标上都优于通用情况的发布/订阅通信协议.