A neural-network contention controller for packet switchingnetworks

A novel approach to solving the output contention in packet switching networks with synchronous switching mode is presented. A contention controller has been designed based on the K-winner-take-all neural-network technique with a speedup factor to achieve a real-time computation of a nonblocking switching high-speed high-capacity packet switch without packet loss. Simulation results for evaluation of the performance of the K-winner network controller with 10 neurons are presented to study the constraints of the "frozen state" as well as those of same initial state. An optoelectronic contention controller constructed from a K-winner neural network is proposed.     

Building a peer-to-peer content distribution network with high performance,scalability and robustness

Content distribution networks (CDN) are fundamental, yet expensive technologies for distributing the content of web-servers to large audiences. The P2P model is a perfect match to build a low-cost and scalable CDN infrastructure for popular websites by exploiting the underutilized resources of their user communities. However, building a P2P-based CDN is not a straightforward endeavor. In contrast to traditional CDNs, peers are autonomous and volunteer participants with their own heterogeneous interests that should be taken into account in the design of the P2P system. Moreover, churn rate is much higher than in dedicated CDN infrastructures, which can easily destabilize the system and severely degrade the performance. Finally and foremostly, while many P2P systems abstract any topological information about the underlying network, a top priority of a CDN is to incorporate locality-awareness in query routing in order to locate close-by content. This paper aims at building a P2P CDN with high performance, scalability and robustness. Our proposed protocols combine DHT efficiency with gossip robustness and take into account the interests and localities of peers. In short, Flower-CDN provides a hybrid and locality-aware routing infrastructure for user queries. PetalUp-CDN is a highly scalable version of Flower-CDN that dynamically adapts to variable rates of participation and prevent overload situations. In addition, we ensure the robustness of our P2P CDN via low-cost maintenance protocols that can detect and recover from churn and dynamicity. Our extensive performance evaluation shows that our protocols yield high performance gains under both static and highly dynamic environments. Furthermore, they incur acceptable and tunable overhead. Finally we provide main guidelines to deploy Flower-CDN for the public use.     

通信网络设计:传输容量、设计代价和可扩展性

虽然目前围绕复杂网络的流量动力学研究日益增多,但这类研究尚存在一些共性问题.例如:所采用的网络流模型过于简单;没有考虑网络设计的代价和可扩展性.本文建立在一个更为真实的网络流模型之上,同时为网络设计问题提供了更全面的视角.本文在广为研究的传输容量之外又引入了另外两个设计目标:网络设计代价和可扩展性.通过引入笛卡尔坐标系统,本文量化分析了不同的网络设计因素是如何独立地影响以及通过相互作用共同影响这些网络设计目标的.结果表明,不同的网络拓扑结构呈现出不同的网络设计可行解空间形状,并在实现高效费比和可扩展的网络设计目标上表现出不同的能力.特别地,文章指出现有的经验性网络设计背后隐藏的理念不能满足高效费比和可扩展的网络设计需求,并提出了针对BA网络的一种高效费比和可扩展的设计方案,即有效路由结合基于有效介数的链路带宽分配方案.而如果从头开始设计一个全新的网络,则随机网络是达成上述目标的最佳选择.     

Measurement and performance of a cognitive packet network

As the size of the Internet grows by orders of magnitude both in terms of users, number of IP addresses, and number of routers, and as the links we use (be they wired, optical or wireless) continuously evolve and provide varying reliability and quality of service, the IP based network architecture that we know so well will have to evolve and change. Both scalability and QoS have become key issues. We are currently conducting a research project that revisits the IP routing architecture issues and proposes new designs for routers. As part of this effort, this paper discusses a packet network architecture called a cognitive packet network (CPN), in which intelligent capabilities for routing and flow control are moved towards the packets, rather than being concentrated in the nodes. In this paper we outline the design of the CPN architecture, and discuss the quality-of-service based routing algorithm that we have designed and implemented. We then present our test-bed and report on extensive measurement experiments that we have conducted.     

Communication network designing:Transmission capacity,cost and scalability

Despite the growing number of works centering around the traffic dynamics on complex networks,these researches still have some common shortcomings,e.g.,too simple traffic flow model and lack of considerations for the designing cost and scalability issues.This paper builds on a more realistic traffic flow model,and offers a holistic view on the network designing problem.In addition to the extensively studied transmission capacity,this paper takes designing cost and scalability as two other designing objectives,and presents a quantitative study of how different designing choices independently and collectively influence these objectives by the introduction of a cartesian coordinate system.It is shown that different kinds of network topologies display different shapes of achievable solution spaces and exhibit different abilities to achieve cost-effective and scalable designing.In particular,we find the philosophy underlying empirical network designing and engineering today fails to meet the cost-effective and scalable designing requirements,and propose a cost-effective and scalable designing scheme for BA-like networks,i.e.,the efficient routing combined with effective betweenness based link bandwidth allocation.In addition,when designing a thoroughly new network from the beginning,we find that ER network is a good candidate to achieve cost-effective and scalable designing in most settings.     

Wavelength converter sharing in a WDM optical packet switch: dimensioning and performance issues

This paper analyses the performance of an optical packet switch making use of fiber delay-lines to resolve output packet contentions and equipped with a set of tunable wavelength converters (TOWCs) shared among the input lines. The wavelength converters can shift optical packets to any wavelength of the output link which they are directed to. An analytical model is developed to evaluate the number of TOWCs needed to satisfy given performance requirements. Moreover, a sensitivity analysis with respect to both the optical packet switch parameters and the traffic load is carried out. The obtained results show that the number of converters needed to maintain the packet loss probability is much lower than that needed by the switches proposed in literature.     

Tiny Tera: a packet switch core

Describes Tiny Tera: a small, high-bandwidth, single-stage switch. Tiny Tera has 32 ports switching fixed-size packets, each operating at over 10 Gbps (approximately the Sonet OC-192e rate, a telecom standard for system interconnects). The switch distinguishes four classes of traffic and includes efficient support for multicasting. We aim to demonstrate that it is possible to use currently available CMOS technology to build this compact switch with an aggregate bandwidth of approximately 1 terabit per second and a central hub no larger than a can of soda. Such a switch could serve as a core for an ATM switch or an Internet router. Tiny Tera is an input-buffered switch, which makes it the highest bandwidth switch possible given a particular CMOS and memory technology. The switch consists of three logical elements: ports, a central crossbar switch, and a central scheduler. It queues packets at a port on entry and optionally prior to exit. The scheduler, which has a map of each port's queue occupancy, determines the crossbar configuration every packet time slot. Input queueing, parallelism, and tight integration are the keys to such a high-bandwidth switch. Input queueing reduces the memory bandwidth requirements: When a switch queues packets at the input, the buffer memories need run no faster than the line rate. Thus, there is no need for the speedup required in output-queued switches     

A four-terabit packet switch supporting long round-trip times

This 4-TBPS packet switch uses a combined input- and crosspoint-queued (CICQ) structure with virtual output queuing at the ingress to achieve the scalability of input-buffered switches, the performance of output-buffered switches, and low latency.     

A new shared-buffer packet switch in ATM networks

To improve overall cell loss rate and fairness problem in shared-buffer switches, a new output-queued shared-buffer switch is proposed in this paper. In the proposed shared-buffer switches, a novel structure of output queues is used to improve the behavior of cell loss rate and resolve the fairness problem. Performance of the proposed output-queued shared-buffer switch is analyzed and compared. According to the analyzed results, the proposed shared-buffer switches show superior performance and overcome the fairness problem. Besides, the proposed buffer architecture is simple to implement in the high-speed packet switches.     

A scheduler ASIC for a programmable packet switch

While the Internet is successful in supporting traditional data-only traffic, an integrated services Internet is inevitable with the emergence of new applications such as voice, video, multimedia, and interactive video conferencing. Such an integrated services network should support a wide range of applications with diverse quality of service requirements and traffic characteristics. Provision for quality of service in packet networks in general, and in the Internet in particular, is the focus of most of the recent developments in switching and routing system design. We designed a generic, single-queue scheduler engine for use in a programmable packet switch/router to handle IP packets, ATM cells, or a combination of both. Comprising 275,000 gates, the 0.35-micron ASIC is incorporated into a prototype programmable packet switch     

A new linear adaptive controller: design, analysis and performance

The certainty equivalence and polynomial approach, widely used for designing adaptive controllers, leads to “simple” adaptive control designs that guarantee stability, asymptotic error convergence, and robustness, but not necessarily good transient performance. Backstepping and tuning functions techniques, on the other hand, are used to design adaptive controllers that guarantee stability and good transient performance at the expense of a highly nonlinear controller. In this paper, we use elements from both design approaches to develop a new certainty equivalence based adaptive controller by combining backstepping based control law with a normalized adaptive law. The new adaptive controller guarantees stability and performance, as well as parametric robustness for the nonadaptive controller, that are comparable with the tuning functions scheme, without the use of higher order nonlinearities     

PF_RING与NAPI结合的捕包性能优化和仿真

频繁中断响应、冗余的数据拷贝和上下文切换等是影响网络数据包捕获性能的主要因素。为了减少这些因素的影响,提出将PF_RING与NAPI结合应用到捕包过程,以对性能进行整体优化。比较了PF_RING与传统数据包捕获机制的差异,分析了两者结合的优势,搭建实验平台,采用内核发包形式,进行实验仿真。在仿真实验中,从捕包率和处理效率与传统方式进行比较,分析实验数据得出该方法可以有效地提高捕包性能。     

A gradual neural-network algorithm for jointly time-slot/code assignment problems in packet radio networks

A gradual neural network (GNN) algorithm is presented for the jointly time-slot/code assignment problem (JTCAP) in a packet radio network in this paper. The goal of this newly defined problem is to find a simultaneous assignment of a time-slot and a code to each communication link, whereas time-slots and codes have been independently assigned in existing algorithms. A time/code division multiple access protocol is adopted for conflict-free communications, where packets are transmitted in repetition of fixed-length time-slots with specific codes. GNN seeks the time-slot/code assignment with the minimum number of time-slots subject to two constraints: (1) the number of codes must not exceed its upper limit and (2) any couple of links within conflict distance must not be assigned to the same time-slot/code pair. The restricted problem for only one code is known to be NP-complete. The performance of GNN is verified through solving 3000 instances with 100-500 nodes and 100-1000 links. The comparison with the lower bound and a greedy algorithm shows the superiority of GNN in terms of the solution quality with the comparable computation time.     

学校计算机网络性能的优化

随着计算机网络的发展,学校使用计算机网络进行教学和办公也变得越来越普遍.如何优化学校计算机网络性能,保证网络能够可靠、安全、高效地运行,是网络管理员普遍关注的一个问题.本文主要从学校计算机网络的构成出发,讲述了网络性能优化的一些做法,希望对读者有所裨益.     

A unified framework for chaotic neural-network approaches to combinatorial optimization.

As an attempt to provide an organized way to study the chaotic structures and their effects in solving combinatorial optimization with chaotic neural networks (CNN), a unifying framework is proposed to serve as a basis where the existing CNN models ran be placed and compared. The key of this proposed framework is the introduction of an extra energy term into the computational energy of the Hopfield model, which takes on different forms for different CNN models, and modifies the original Hopfield energy landscape in various manners. Three CNN models, namely the Chen and Aihara model with self-feedback chaotic simulated annealing [CSA] (1995, 1997), the Wang and Smith model with timestep CSA (1998), and the chaotic noise model, are chosen as examples to show how they can be classified and compared within the proposed framework.     

Rimgnet: A packet switched local network with decentralized control

Ringnet is a commercially available 8 MHz packet switched communication system designed to connect hosts in the same building, usually less than several hundred feet apart, although it can also be used for loosely coupled multiprocessor systems. Control of the network is completely decentralized and depends only on a circulating token for arbitration of the communication medium. This paper describes the hardware and software architecture of the network and discusses some of the design principles used in its implementation.     

A MOS-based dynamic memetic differential evolution algorithm for continuous optimization: a scalability test

Continuous optimization is one of the areas with more activity in the field of heuristic optimization. Many algorithms have been proposed and compared on several benchmarks of functions, with different performance depending on the problems. For this reason, the combination of different search strategies seems desirable to obtain the best performance of each of these approaches. This contribution explores the use of a hybrid memetic algorithm based on the multiple offspring framework. The proposed algorithm combines the explorative/exploitative strength of two heuristic search methods that separately obtain very competitive results. This algorithm has been tested with the benchmark problems and conditions defined for the special issue of the Soft Computing Journal on Scalability of Evolutionary Algorithms and other Metaheuristics for Large Scale Continuous Optimization Problems. The proposed algorithm obtained the best results compared with both its composing algorithms and a set of reference algorithms that were proposed for the special issue.     

Approximate performance modeling and decision making for manufacturing systems: A queueing network optimization framework

In this paper we discuss queueing network methodology as a framework to address issues that arise in the design and planning of discrete manufacturing systems. Our review focuses on three aspects: modeling of manufacturing facilities, performance evaluation and optimization with queueing networks. We describe both open and closed network models and present several examples from the literature illustrating applications of the methodology. We also provide a brief outline of outstanding research issues. The paper is directed towards the practitioner with operations research background and the operations management researcher with interest in this topic.     

Facing scalability: Naming faces in an online social network

Automatically naming faces in online social networks enables us to search for photos and build user face models. We consider two common weakly supervised settings where: (1) users are linked to photos, not to faces and (2) photos are not labeled but part of a user's album. The focus is on algorithms that scale up to an entire online social network. We extensively evaluate different graph-based strategies to label faces in both settings and consider dependencies. We achieve results on a par with a recent multi-person approach, but with 60 times less computation time on a set of 300K weakly labeled faces and 1.4 M faces in user albums. A subset of the faces can be labeled with a speed-up of over three orders of magnitude.