Review on Fault-Tolerant NoC Designs

By benefiting from the development of the semiconductor technology, many-core system-on-chips (SoCs) have been widely used in electronic devices. Network-on-chips (NoCs) can address the massive stress of on-chip communications due to the advantages of high bandwidth, low latency, and good flexibility. Since deep sub-micron era, the reliability has become a critical constraint for integrated circuits. To provide correct data transmission, fault-tolerant NoCs have been researched widely in last decades, and many valuable designs have been proposed. This work introduces and summarizes the state-of-the-art technologies for fault diagnosis and fault recovery in fault-tolerant NoCs. Moreover, this work makes prospects for the future's research.     

一种基于FPGA的新颖的高容错加法器的设计

为了解决传统TMR结构的CMFs失效问题,根据加法器的结构特点提出了操作数循环移位及取反算法(TOIR-SO).此方法相对于传统的TMR结构能够使TMR系统失效率降低47%.同时对逻辑运算的基本单元全加器进行了改进,改进后加法器中任何一个失效只能影响一位"和"结果而不会对其它位产生影响从而进一步提高了加法器的容错能力.     

针对路径故障与局部拥塞的NoC容错路由算法

片上网络作为一种新型片上互连架构,克服了片上系统在发展中遭遇的瓶颈问题。然而,片上网络中的路由器故障以及路由器之间的链路故障都会造成网络性能损失。对此,文章提出一种针对路径故障与局部拥塞的NoC容错路由算法。首先,设计了一种相隔节点间路径故障模型,该模型下的路由器以较小的开销为代价,动态感知两跳以内的路径故障状态。其次,提出了一种新颖的更能准确反映局部网络拥塞状态的拥塞模型来均衡网络流量。最后,当网络无故障时,算法保证走最优路径;有故障时,算法不仅可以实现容错还能保证网络具有良好的性能。实验表明,在无故障的情况下,本文方案相较于对比对象延迟降低了10％～20％,吞吐率提高了25％左右。在有故障的情况下,本文方案较对比对象的优势更加明显。     

具有在线修复能力的强容错三模冗余系统设计及实验研究

 为提高太空恶劣环境中电子系统的可靠性,提出了一种具有芯片级在线修复能力的强容错三模冗余(TMR)系统结构及设计方法,可在不影响系统正常工作的前提下实现故障模块的在线修复.该系统采用TMR结构,可实时检测定位故障模块;模块采用组件备份法设计,故障发生时可通过备件切换法快速自修复,模块中每个故障组件均可通过进化进行修复;并通过异构冗余降低2个以上模块同时故障的概率.以具有片内三模冗余的三阶高密度双极性(HDB3)编码器系统设计为例,对系统结构和各种容错修复机制进行了验证,结果表明系统可靠性得到很大提高.     

Mesh-Based Network Convergence and Cooperation

More and more distinctive heterogeneous features are exhibited in wireless networks in multitude of networks, technologies, terminals, operation and management, and more. Effective convergence and cooperation of heterogeneous networks can be achieved through Mesh technology. The IP-based convergence integrates common features of heterogeneous networks, while the cooperation coordinates personalities of the networks. By using the convergence and cooperation, a standardized integration of separated and localized predominant capabilities and resources of heterogeneous networks can be fulfilled to enable the evolution into an ubiquitous and omnipotent intelligent network.     

QoS Routing for Mesh-Based Wireless LANs

Wireless LANs with their increased data rate become an attractive technology for connecting mobile users to the Internet. Efficient deployment of wireless LANs will require the ability to extend the wireless LANs coverage without the need to deploy a very large number of access points. Mesh-based wireless LANs are an attractive solution to this problem. These networks extend wireless LAN coverage by using each node in the network as a router, resulting in a multihop topology. In this paper we introduce a novel routing algorithm, wireless mesh routing (WMR), that provides quality-of-service (QoS) support and accounts for the characteristics of both infrastructure-based wireless LANs and ad hoc networks. The algorithm is validated using the OPNET modeler. The simulation results show that the mesh network using WMR protocol can provide QoS support and react dynamically to the network status changes with low control overheads. Using the proposed software architecture the proposed routing protocol can be implemented with any MAC protocol, resulting in easy implementation in existing wireless nodes.     

PRIME: Peer-to-Peer Receiver-Driven Mesh-Based Streaming

The success of file swarming mechanisms such as BitTorrent has motivated a new approach for scalable streaming of live content that we call mesh-based Peer-to-Peer (P2P) streaming. In this approach, participating end-systems (or peers) form a randomly connected mesh and incorporate swarming content delivery to stream live content. Despite the growing popularity of this approach, neither the fundamental design tradeoffs nor the basic performance bottlenecks in mesh-based P2P streaming are well understood.      

基于通讯的NoC设计

近年来，一种全新的集成电路体系结构——Network on Chip（NoC）已经成为徽电子学科研究的热点佃题之一，其核心思想是将计算机网络技术移植到芯片设计中来，从体系结构上彻底解决片上通讯的瓶颈问题。文章提出了一种基于通讯的NoC设计方法，通过监控和协调NoC的网络通讯来获得更好的性能．并总结了实现该设计方法所必须研究的关键技术。     

一种基于SystemC的片上网络建模与仿真方法

随着芯片上集成度的提高,设计复杂性和时钟同步的问题也越来越严重,因此,面向通讯为中心的片上网络(NoC)将成为未来片上互连的主要方式.由此建立了一种可对路由内部单元细化配置的片上网络模拟器:采用分组数据的虫孔(Wormhole)交换策略,使用仿真速度较快的SystemC事务级语言搭建了片上网络模型.同时提出了一种较为精确的仿真方法,模拟结果显示能适用于多核结构的片上网络的模拟仿真.     

具有容错策略的波分复用光传输网络

针对波分复用光传输网络,提出了一种关键光路的容错策略。这种策略为网络中的每个节点预保留信道资源。在网络出现故障后,不同节点之间预保留的信道可以共享,极大地减少了预保留的资源数量。实验模拟显示,在单个错误恢复率略略降低(不到1%)的前提下,预保留的资源只占总资源的不到18%,恢复时间没有延长。结合这种容错策略,也提出了新的分布式容错策略,恢复时只需要非常少的额外资源,而网络能够快速地恢复多个错误。     

Redundancy averaging with large arrays

Redundancy averaging is a spatial averaging technique that can lead to improved spatial correlation matrix estimates in terms of reduced intersignal coherence and reduced variance, but it has the undesirable characteristic of potentially biasing direction estimates. It is shown that for linear equally spaced arrays, any basis induced in the direction estimates by redundancy averaging vanishes as the number of sensors becomes large     

Mesh结构电源/地布线网络分析方法综述

在深亚微米VLSI设计中，获得电源／地布线网络优化设计的关键是在设计周期里对其进行有效的分析。文章在叙述Mesh结构电源／地布线网络分析模型的基础上，简要介绍了电源／地布线网络分析方法的研究进展，并指出了该研究领域存在的一些问题。     

Fault-Tolerant Software

Limitations in the current capabilities for verifying programs by formal proof or by exhaustive testing have led to the investigation of fault-tolerance techniques for applications where the consequence of failure is particularly severe. Two current approaches, N-version programming and the recovery block, are described. A critical feature in the latter is the acceptance test, and a number of useful techniques for constructing these are presented. A system model for the recovery block is introduced, and conclusions derived from this model that affect the design of fault-tolerant software are discussed.     

低功耗优先的片上网络映射优化方法

在分析片上网络通讯功耗与通讯流量的关系模型的基础上,针对片上网络设计中的映射问题,提出了一种新的降低通讯功耗的映射方法,该方法首先对映射过程做预处理,划分成若干候选图,将通讯量大的IP核映射到与其他资源节点距离较短的位置上,利用预处理的结果产生初始解,结合流量估算技术对映射空间动态搜索,从而实现将通讯任务图中的IP核映射到NoC结构图的资源节点上.实验结果表明该方法能有效地降低NoC的通讯流量,从而更加适合求解片上网络的低功耗映射问题.     

基于SystemC的片上网络全系统模拟器

设计了一个基于SystemC的可配置片上网络(NoC)全系统模拟器.该模拟器实现了路由器、处理节点、存储节点及IO节点的周期精确的事务级建模,同时具备分析各路由器节点性能及影响因素的功能.实验结果表明该模拟器具有快速、灵活及准确性高的优点,在2.4GHz的Intel Core2处理器上对4×4的NoC进行周期精确的仿真约为130kcycles/s,测试程序在该模拟器运行的结果与系统实现后的实际结果相差小于4%.     

具有扩展局部连通性超立方体中的容错路由

在局部连通性的基础上,提出了针对超立方体网络Hn的扩展的局部k-维子立方体连通性概念,证明了具有扩展的局部k-维子立方体连通性的Hn中正确结点问是连通的;提出了超立方体网络Hn中基于扩展局部k-堆子立方体连通性的路由算法。     

On Standby Redundancy with Delayed Repair

This paper derives a) the Laplace-Stieltjes transform of the time-to-system-failure distribution, and b) the mean time-to-system-failure. The system consists of several elements with one repair facility which remains idle until a queue of failed units is built up.     

基于NoC的容错路由算法

各种各样的软件和硬件上的错误都会破坏网络的数据传输,因此研究No C网络的容错算法是非常必要的。在基于XY路由算法的基础上提出了改进的容错路由算法,当链路或者传输节点之间发生错误时,可通过重新设置路由规则来获取一条有效的路由路径。在FPGA上进行路由容错算法的仿真,并和目前常用的几种路由算法在所适用拓扑、是否防止死锁等方面进行对比。仿真结果显示改进的路由容错算法性能优越,是可行的。     

一种基于多项复杂特征的 Internet路由级拓扑建模方法

通过对CAIDA多监测点获得的测量数据的分析,提取出表征Internet路由级拓扑规律的多项特征值.在特征值分析的基础上,综合考虑了路由级拓扑的动态变化,拓扑的层次性,以及节点间非线性连接的特点,提出了面向Internet路由级拓扑的DNLH(Dynamic Non-Linear Hierarchy)模型,并给出建模算法.以实际数据为依据,结合遗传算法优化了节点间非线性连接参数.通过实验分析,证明了该建模方法的有效性.     

Functional Test of Mesh-Based NoCs with Deterministic Routing: Integrating the Test of Interconnects and Routers

In this work, a comprehensive functional-based test method is presented to integrate the test of Network-on-Chip interconnects and routers. Experimental results show that fault coverage can reach up to 100% of interconnect faults and over 90% of router faults. The test structures needed to implement the test method are presented and the scalability of the method is discussed. We show that our approach scales well with the number of routers and channel width. The cost of the functional test strategy and the cost of a scan-based strategy are also analyzed in order to present scenarios where each strategy fits better.