一种基于NIC的RDMA可靠传输协议的设计与实现

高性能计算机不断增长的规模和复杂性使得可靠性成为影响高性能计算机系统可用性的关键因素,系统互连网络是高性能计算机的重要组成部分,其可靠性是高性能计算机系统设计必须考虑的重要问题。针对高性能计算机系统互连网络可能出现的故障,提出一种基于NIC实现的RDMA可靠传输协议,给出了一种通用的设计实现方案,并对该方案的几种具体优化设计实现方法进行了讨论。提出的可靠传输协议及实现方案能容忍系统互连网络可能出现的多种网络故障,并能尽量减少实现可靠传输所带来的额外开销。实验结果表明,所提出的RDMA可靠传输的实际测试性能与无连接RDMA传输相当。     

基于补偿的无线传感器网络可靠性数据传输协议

该文在无线传感器网络分簇路由的基础上,提出了一种基于补偿机制的可靠数据传输协议PCI,该协议可以实现高效、及时、可靠的数据收集和传输。PCI主要包括三个重要部分:1)可靠性概率计算机制;2)消息分类机制;3)智能均衡机制。通过调用这三种机制,PCI协议可以减少冗余信息以提高传输性能,并且可以补偿信息发送量的不足以保证数据传输的可靠性。通过仿真可以发现,PCI协议比LEACH和ECDG有着更快的数据采集速率并且使整个网络具有更好的稳定性。     

实时大规模远程实验通信方案研究

针对开放式远程实验平台的高并发、实时性、可靠性和安全性需求,设计并实现了一种高并发访问的远程实验通信方案。该方案能够实时连接大规模传感器网络,具有实验仪器与用户并发访问、实时可靠消息传输、网络安全控制和仪器安全保障功能,能有效提高远程实验的通信性能。在模拟大规模用户实时并发访问的情况下进行性能测试,测试结果表明该方案不仅能满足实验通信的并发需求,还能有效确保消息的实时可靠传输和安全控制,具有较高的应用价值。     

面向安全应用消息QoS的接入协议研究

为了实现车联网安全应用消息的低时延高可靠传输,设计了一种基于时分多址(time division multiple access,TDMA)面向安全应用消息服务质量(quality of service,QoS)的多址接入协议QoS-oriented TDMA。该协议采用TDMA技术实现,并同时考虑了协同感知信息(cooperative awareness message,CAM)和分布式环境通知消息(decentralized environmental notification message,DENM)两种安全应用消息的传输需求。采用交通仿真软件SUMO和网络仿真软件OMNe T++相结合的,交通与网络仿真双向耦合、实时交互的车联网仿真平台Veins,对QoS-oriented TDMA协议进行了仿真实现和性能验证。仿真结果表明,CAM传输可靠性虽然没有明显改善,但是QoS-oriented TDMA协议提高了DENM传输可靠性,并降低了CAM和DENM传输时延。与IEEE 802.11p协议相比,QoS-oriented TDMA协议高速场景和市区场景中CAM接收概率分别降低了5.8%和7.1%,传输时延分别降低了11.4%和12.8%;DENM的接收概率分别提高了10.5%和12.9%,传输时延分别降低了13.3%和15.1%。     

基于UDP的报文可靠传输技术研究

为了在较小的网络开销下保证数据的可靠传输,本文提出了基于UDP的无连接可靠传输协议LRC,通过对协议支持的传输方式和使用流程进行设计,使其在继承了UDP传输实时性的同时具备了可靠传输的功能,特别是LRC协议支持的单任务可靠发送方式可以获得比较高效的传输性能。最后,对该协议中的两种可靠发送方式进行性能分析。     

基于MOM的异步可靠Web服务模型的设计与实现

提出了一个基于消息中间件(MOM)的异步可靠Web服务模型,该模型分为服务器端和客户端两大部分,支持同步和异步调用,解决了Web服务调用中可能会出现的异常问题,如服务短暂停止和网络出现故障等.对该模型进行了分析,然后对该模型的客户端和服务器端进行了设计实现.最后该模型通过消息持久化、消息队列和消息状态以及优先级管理和消息保证迭达等机制保证了消息传输的可靠性,提高了Web服务的服务质量.     

GSRP--一种支持连接恢复和会话移动的消息传输协议

从网格用户环境等场景和移动网络特点出发,归纳出网络协议中持久性、全局性、松耦合、内容和传输分离的原则,并提出支持连接恢复和会话移动的网格服务请求协议(GSRP)．GSRP采用松耦合的应用会话层和消息传输层两层结构,分别负责消息内容及应用之间消息交互和消息封包及传输．全局性的持久会话句柄为不同主机和不同应用之间实现网络会话的协同和移动提供了开发接口,软连接在持久会话句柄和底层网络链路之间建立映射,完成消息的传输．单边操作和延迟容忍技术被用在数据通信和状态管理中,简化了应用开发和维护．给出GSRP的设计和实现,结果表明GSRP在没有增加大的额外开销的同时,获得了休眠、移动和对底层网络的适应能力,可以满足所关心的新应用的需求．     

面向安全应用消息传输的异构网络选择算法

为了实现异构车联网中安全应用消息的低时延高可靠传输,针对基于DSRC/LTE的异构车联网,设计了一种面向安全应用消息传输的网络选择算法。该算法从安全应用消息传输服务质量(quality of service,QoS)需求出发,综合考虑了安全应用消息传输QoS的多项参数指标,包括端到端时延、丢包率、传输速率。算法首先利用层次分析法计算出安全应用消息传输QoS各个参数的权重值,然后分析了各个候选网络支持安全应用消息传输QoS的能力,最后确定了安全应用消息传输QoS的满意度。采用双向耦合、实时交互的车联网仿真平台Veins LTE对面向安全应用消息传输QoS的网络选择算法进行了仿真实现和性能验证。仿真结果表明,面向安全应用消息传输的网络选择算法可以根据安全应用消息传输的不同QoS需求,合理地分配网络资源。一方面,与仅使用已经实现的QoS-oriented TDMA协议相比,同时使用QoS-oriented TDMA协议和面向安全应用消息传输的网络选择算法的异构车联网,在高速场景和市区场景中安全应用消息的接收概率分别提高了13.7%和10.4%,传输时延分别减少了8.6%和11.5%。另一方面,面向安全应用消息传输的网络选择算法在高速场景和市区场景中,与匹配博弈网络选择算法相比,安全应用消息的接收概率分别提高了11.6%和7.5%,传输时延分别减少了13.2%和15.3%。     

基于带宽管理的实时流媒体传输策略

流控制传输协议（SCTP）是基于不可靠的、无连接的分组IP网络上的一种面向消息的、可靠的传输层协议,它更适合在较高的网络带宽环境下的应用。而作为一种部分可靠版本的PRSCTP在带宽受限的网络环境下有更好的性能表现。该文提出一种适合实时流媒体网络应用的方案BMSCTP,它通过带宽估计机制,实现自适应地在SCTP和PRSCTP之间进行切换。出于性能分析和评估的需要,设计了一组仿真实验。实验结果显示,动态网络环境下传输实时流媒体时,在投递率和传输延迟等性能指标方面,BMSCTP要优于SCTP和PRSCTP。     

一种新的效用冗余混合DTN路由算法

针对容迟网络(DTN)的高延迟、间歇性连接和资源受限等特点,提出一种新的效用冗余混合DTN路由算法NHRA。根据节点的相遇频率和连通时长等信息估算消息转发效用,指定消息在网络中的最大拷贝数,根据消息转发效用结合基于二叉树的方法来分配消息拷贝任务,实现消息的多路径并行传输,在提高传输性能的同时,有效地控制了网络开销。仿真实验结果表明,与现有相关算法相比,该算法可获得更高的消息传达率,并且花费的通信开销更低。     

无线/移动网络中基于价格的接纳控制方案

无线/移动网络中,为降低切换连接丢弃率,通常以新连接请求阻塞率的增加及网络资源利用率的降低为代价。提出一种基于价格的接纳控制方案,利用价格杠杆调节用户对网络资源的使用,从而达到提高系统的连接级QoS水平,并增加网络利润的目的。仿真实验证明了这一结论。     

具有高可靠特征的无线虚拟网络映射方法研究

针对虚拟网络映射过程中链路资源受限和网络生存性问题，分别提出一种基于节点扩展资源的节点映射方案和一种虚拟网络重映射方案。为克服节点映射后链路映射阶段的资源不足问题，通过将节点扩展资源大的节点作为承载节点以确保链路映射的可靠性和高效性，链路映射阶段采用一种基于最小链路代价的映射方案以获得高可靠映射，在此基础上针对网络故障采取一种重映射策略，将失效节点迅速重映射至候选节点集中以确保虚拟网络服务的连续性。通过仿真对几种算法性能进行对比，结果表明所提算法在虚拟网络映射成功率、虚拟网络恢复成功率和资源负载利用率等方面均取得较好结果。     

Particle swarm optimization-based DRWA for wavelength continuous WDM optical networks using a novel fitness function

Greater demand of bandwidth and network usage flexibility from customers along with new automated means for network resource management has led to the concept of dynamic resource provisioning in WDM optical networks where unlike the traditional static channel assignment process, network resources can be assigned dynamically. This paper examines a novel particle swarm optimization (PSO)-based scheme to solve dynamic routing and wavelength assignment (dynamic RWA) process needed to provision optical channels for wavelength continuous Wavelength Division Multiplexed (WDM) optical network without any wavelength conversion capability. The proposed PSO scheme employs a novel fitness function which is used during quantization of solutions represented by respective particles of the swarm. The proposed fitness function takes into account the normalized path length of the chosen route and the normalized number of free wavelengths available over the whole route, enabling the PSO-based scheme to be self-tuning by minimizing the need to have a dynamic algorithmic parameter ‘α’ needed for better performance in terms of blocking probability of the connection requests. Simulation results show better performance of the proposed PSO scheme employing novel fitness function for solving dynamic RWA problem, not only in terms of connection blocking probability but also route computation time as compared to other evolutionary schemes like genetic algorithms.     

基于网络编码的分簇传感器网络链路容错策略

许多紧急事件的应用对无线传感器网络的数据传输的可靠性有比较高的要求, 因此对传感器网络数据传输的容错性研究十分重要。传统的1+1数据保护的容错模式带来的额外负载较大地增加了传感器节点的能耗。提出了基于网络编码的分簇结构传感器网络链路容错策略,不仅可以实现对链路的容错,而且可以降低网络中转发的数据包数量,从而降低传感器节点的能耗。     

WSN中基于一次预共享密钥的安全管理方案

无线传感器网络在部署初始阶段,由于未被恶意者察觉或者破解存在难度,节点通常能够在一定时间内具有抵抗劫持的能力。该文基于此分析提出一种适合于静态网络的基于一次性预共享密钥的安全管理方案SMOPK,方案在无线传感器网络部署的初始阶段利用一次性预共享密钥进行密钥协商等,从而在相邻节点间建立保密的安全连接,建立与实际拓扑一致的安全拓扑。分析表明SMOPK方案具有开销小安全性强的特点,适合于静态无线传感器网络的安全要求。     

Deadline-based connection setup in wavelength-routed WDM networks

This article addresses the ubiquitous topic of quality of service (QoS) aware connection provisioning in wavelength-routed WDM optical networks. The impact of the connection setup time of an optical connection has not been adequately addressed in the open literature. As such, this paper presents a novel approach that uses the optical connection setup time as a service differentiator during connection provisioning. The proposed approach utilizes the Earliest Deadline First (EDF) queueing algorithm to achieve deadline-based connection setup management with the deadline being the setup time requirement of an optical connection. The proposed EDF-based approach would allow the network operator to improve the QoS perceived by the end clients. Performance of this novel scheme is analyzed by accurately calculating various parameters, such as the fraction of connections provisioned on-time (i.e. prior to deadline expiration) and the average time it takes to successfully setup a connection. In addition, the presented approach is validated by a simulation that analyzes the performance of the proposed connection setup scheme in the specific context of the National Science Foundation Network (NSFNET). The obtained results show that a deadline-based setup strategy can minimize blocking probability while achieving QoS differentiation.     

Priority-enabled optical shared protection: An online efficiency evaluation study

The availability of an optical connection is considered to be a critical service differentiator in WDM optical networks. In this regard, the design of a protection scheme that is able to improve the availability of high priority connections while making efficient use of optical resources is a major concern for optical network operators. In a previous work, we proposed the so-called priority-aware shared protection as a potential survivability scheme for next generation WDM networks to deal with the previously exhibited concern.

This paper develops an online study whose main purpose is to assess the efficiency of the aforementioned protection scheme. Through this study, we show that the priority-aware shared protection strategy is able to achieve both the best efficiency in terms of resource usage and in terms of availability satisfaction rate compared to existing protection solutions.     

Dynamic rerouting of guaranteed Quality-of-Service connections

To support the widespread demands for integrated services various solutions have been proposed to provide guaranteed Quality-of-Service (QoS) services in packetswitched networks. These solutions usually require fixed route selection and resource reservation on a per connection basis. The routing and reservation decisions, combined with load fluctuations, introduce the problems of network unavailability and loss of network management flexibility. We believe that these problems can be alleviated by transparently rerouting connections to properly balance the network load. In this paper, we present a solution that supportstransparent route modification ofguaranteed QoS connections during the life time of the connections. As the emphasis of this work is to support transparency we must ensure that during and after the rerouting, the QoS guarantees made to the client are preserved, and the interface instances at both the sender and receiver are unaltered. Our solution includes the mechanisms that support the transparent rerouting of connections, and a routing algorithm that is based on the traffic characteristics, the delay, and the jitter requirements of a connection. This routing algorithm can also use various administrative constraints and facilitate the reuse of resources currently reserved by the connection. To optimize the solution, additional mechanisms have been introduced to reduce the time needed to reroute a connection and to support rerouting of segments of a connection. The solution has been implemented in the context of the Tenet scheme and route changes can be observed and controlled by using the Simple Network Management Protocol (SNMPv 1). Simulation experiments and some initial measurements of the implementation demonstrate that such a scheme is useful, practical, and can be achieved with load overhead.     

Priority-aware optical shared protection: An offline evaluation study

The availability of an optical connection is considered to be a critical service differentiator in WDM optical networks. In this regard, the design of a protection scheme that improves the availability of high priority optical connections and makes efficient use of optical resources is a major challenge faced by optical network operators. In a previous study, we proposed the so-called priority-aware shared protection survivability scheme as a potential solution to this design issue.In this paper, we complement our previous study. More specifically, we develop an offline study whose main purpose is to assess the efficiency of the priority-aware shared protection scheme. Through this study, we show that the priority-aware shared protection strategy as opposed to existing protection strategies is able to achieve the best tradeoff between optical resource usage and optical connections’ availability satisfaction.