UCL Logo Networks Research Group Logo

UCL >> Department of Computer Science >> Networks Research Group >> Seminars
People Overview Projects Publications Seminars Calls for Papers Reading Group SIGs Networking Resources PhD and Job Applications




The Networks Group meets regularly throughout the year for informal discussions, presentations, and a reading group. This is an archive of the seminar abstracts and links to the seminar slides for 2003.

January 2004

  • January 14 (Room 214) (first week of term)
    Speaker:Jon Crowcroft (Computer Lab, Cambridge)
    Title: Scalable Ubiquitous Computing Systems slides (ppt)
    Abstract: We are all acquainted with the Personal Computer. We are less aware of the burgeoning numbers of invisible, embedded computers around us in the fabric of our homes, shops, vehicles and even farms.

    They help us command, control, communicate, entertain, and commerce, and these invisible computers are far, far more numerous than their desktop cousins.

    The visible face of computing, the ubiquitous PC, is nowadays generally networked. To date, embedded computing systems have been largely used to replace analog control systems (for reasons of price, performance and reliability). Increasingly, however, we will find systems are integrated into a whole.

    This will lead to a challenge for Computer Science in the form of system complexity. Complexity is at the core of the skill-set of computer science and engineering, but it is also becoming a key piece of the formalisms used to understand other systems in the natural world, in ecology and biology and in physics. With the Internet as large and organic as it already is, we see a complex set of interactions with graph theory, control theory, economics and game theory, and a number of other disciplines being bought to bear and even extended to understand its behaviour. We also see a set of engineering rules of thumb maturing into design principles, which can be applied to other systems.

    Some principles already established in the world of Internet-scale engineering give us hope that we can build systems early (and there are many Ubiquitous Computing projects underway in the UK, EU and world today), with some hope that they will work. However, as systems grow, new problems for performance (stability, availability, etc) will emerge. Critical new areas for concern are the control of multiple resources (scheduling for battery life, randomising timing of events to avoid correlated overload, statistical failure tolerance in very large scale sensor systems). Within the timescales of this challenge, components will even start to draw resources (power) directly from their environment (ambient heat, RF etc), and this has hidden consequences (radio opacity in unusual places for example). The more we look at how such systems will be built, the more we see them vanish into the substance (and ether) around us!

    The core of this challenge then, is to abstract out these engineering design principles, and this will be achieved largely through a process of ``build and learn''. This is a natural complement and sister to the challenge to uncover the Science for Global Ubiquitous Computing, which will have descriptive power. We will have prescriptive solutions (patterns) for the mixed reality environment that will form the next phase of development of cyberspace.

  • January 21
    Speaker:Martin Dodge (CASA, UCL)
    Title: See Inside The Cloud : Some Ways to Map the Internet (slides pdf)
    What does the internet look like? Conventionally, engineers have represented it as a cloud, a useful graphic shorthand to mask its complexity. In my presentation I consider how cartographic maps and graph visualisations are used to represent whats inside the internet cloud. Maps are powerful because they do not just represent space, they are also active in the construction of space inside peoples heads. This is especially so in the construction of peoples cognitive conceptions of the internet, as the infrastructure is largely invisible and intangible in everyday life.

    Over the last thirty years or so, a huge range of different maps of the internet have been produced, with diverse forms and function, from simple geographic plans of cable routes to complex real-time 3D visualisations. They have been produced for a number of distinct purposes from planning network deployment, operational management, to prove academic theories, as grad student projects, for market research, for setting policy and monitoring outcomes, and to try to sell things. And, of course, many have been motivated to map the internet for no particular reason other than because it is there. There are many different aspects of the internet that have been mapped from physical infrastructure, logical layers and protocols, traffic flows, user demographics. The maps cover a range of different scales from individuals, single buildings up to global scale. Many of these maps are beautiful and many more are really rather ugly. A few are actually quite useful, but many more are not very helpful at all. However, all the maps provide a fascinating picture of what the internet looks like, or rather they provide some insights into what people think the internet should look, once the clouds have cleared.

  • January 28
    Speaker:Joe Touch (ISI)
    Title: Virtual Internet Research at the Postel Center (slides(ppt) )
    Abstract: Virtual Internets (VIs) are emerging as a useful tool for managing shared testbed infrastructure, as well as supporting emerging protocols and systems. This talk presents the principles of our Virtual Internet Architecture, and examines its impact on the architecture of end systems, routers, and protocols. The talk also summarizes related research exploring the capabilities of VIs and augmenting system capabilities to support VIs. This includes: the X-Bone system for automated VI deployment and management; the DynaBone system for fault-tolerance and performance via multi-layer virtualization; the NetFS system for providing compartmentalized configuration of network resources; and the DataRouter system for supporting application-directed peer networks via a network-layer string rewriting mechanism.

    Time permitting, we will also demonstrate TetherNet (adapted VI technology), a system for Internet subnet leasing that can undo the effects of NATs, and present a brief summary of our emerging work in designing components to support an all-optical Internet router.

    Bio: Joseph D. Touch is Director of the Postel Center in the Computer Networks Division of USC's Information Sciences Institute (ISI). He received a B.S. with Honors in biophysics and CS from the University of Scranton in 1985, an M.S. in CS from Cornell University in 1987, and a Ph.D. in CS from the University of Pennsylvania in 1992. He joined ISI in in 1992, and his current projects include automatic networks, virtual networks (NetFS, X-Tend), and optical Internets. His research interests include Internet protocols, network architecture, high-speed & low-latency nets, network device design, and experimental network analysis. His past projects range from gigabit LANs (Atomic-2), NIC design (PC-Atomic), and multicast web caching (LSAM), to overlay networks (X-Bone, Dynabone). He is a research associate professor in USC's Departments of Computer Science and Electrical Engineering, where he runs a summer internship program (SGREP) and advises a number of graduate students. He has published over 50 papers and is co-author of a high-speed networks book. Joe is a member of Sigma Xi, IEEE, and ACM. He co-Chairs the IEEE ITC (Internet) committee, is ACM SIGCOMM’s Conference Coordinator, and is active in the IETF. He also serves on the editorial boards of IEEE Network and Computer Networks.


February 2004

  • February 04
    Speaker:Luigi Rizzo
    Title: Overview of the CoMo system (work in progress) (slides)
    Abstract:CoMo (Continuous Monitoring) is a system for capturing Internet traffic at multi-Gbps speeds, and exporting both relevant metrics computed in real time, and the last 24 hours of packet level traces, to a centralized collection system. CoMo would allow network operators to quickly localize and react to undesired network events(e.g., denial of service attacks or router misconfigurations) and, also, to post-process the captured data traffic to further investigate unusual traffic patterns.

    In this talk we will focus on the design challenges posed by asystem of this kind (and basically, by most high speed packet processing systems) using off-the-shelf hardware, in particular with respect to the handling of overloads. We will also present design of the CoMo software and the approach we followed to make it easily extensible.

  • February 11
    Speaker:Soren-Aksel Sorensen
    Title: More than brute force: Petaflops in the modern world. (slides tar.gz)
    Abstract:The desktop revolution has placed massive processing power at the finger tip of most people. Few people today can complain that they have insufficient access to processing power. Ubiquitous computing is seen as a reasonable challenge and we can now see the outline of a processing society similar to those found in science fiction novels. However, High Performance Computing(HPC) seems to have missed the boat. The traditional time shared batch submission environment still rules.

    I will be talking about my vision of HPC and the changes I believe we have to make in order to get to that vision. The challenges are both in the way we build models and in the way we use the available resources. Grid computing can bring us some way towards this goal, but there are many challenges to vistually every branch of computing if we are to succeed.

  • February 18 (Reading week)
    Speaker:Piers O'Hanlon
    Title: Streaming video over multiple simulateous TCP connections. (slides (pdf))
    Abstract: We present a brief investigation into streaming a video source across multiple TCP connections to improve throughput through firewalls and NAT boxes.

  • February 25
    Speaker:Theodore Hong (Imperial College)
    Title: Current developments in Freenet
    Abstract:Freenet is a peer-to-peer network designed to provide efficient and anonymous information publication. It is one of themost widely used such systems, with over two million downloads. Inthis talk, we give an overview of the state of the Freenet network today and discuss some of the new developments currently being experimented with, notably proposed changes to the routing algorithm and mechanisms for load limiting.

  • [Top]

March 2004

  • March 03
    Speakers:John McCarthy, Dimitrios Miras and Angela Sasse
    Title: Sharp or Smooth? Comparing the effects of quantization vs. frame rate for streamed video slides (ppt)
    Abstract: Using a new methodology to evaluate the perceived video quality, we test the claim that high frame rate is more important than quantization when watching high motion video, such as sports coverage. In two studies we examine the relationship between physical quality and perceived quality metrics. In Study 1, 41 soccer fans viewed CIF-sized images on a desktop computer. Study 2 repeated the experiment with 37 soccer fans, viewing the same content, in QCIF size, on a palmtop device. Contrary to existing guidelines, we found that users prefer high-resolution images to high frame rate. We conclude that the rule "high motion = high framerate" does not apply to small screens. With small screen devices,reducing quantization removes important information about the players and the ball. These findings have important implications for service providers and designers of streamed video applications.

  • March 10
    Location: Room 214
    Speaker:Stephano Street
    Title:GDM and JPS systems for distributed computing.
    Abstract: The increasing demand on computational resources has led to much interest in Grid technologies. Among the user areas with such needs is Bioinformatics, which has a seemingly persistent demand for computational cycles. The JPortal & GridDM systems are being designed to help provide a transparent use of resources over multiple domains and application types. The main problems associated with sharing cluster resources over domains is scheduler communication; at the core of the GridDM is a solution to this problem. The JPortal is an application driven single point of entry to the system for registered users; it provides a host of features that help to overcome short falls in the current technologies as discovered in recent research.

  • March 17
    Location: Room 214
    Speaker:Mark Handley
    Title: Internet Denial-of-Service Attacks, slides (pdf)
    Abstract: A discussion of the spectrum of possible denial-of-service attacks onInternet systems, lessons learned, and some random musings onarchitectural changes to reduce the threat. The background material is from this paper Internet Denial of Service Considerations

  • March 24 (last week of term)
    Speaker:Lionel Sacks (EE, UCL)
    Title: "What should a network think about its self?" (slides (pdf))
    Abstract: There have been many initiatives in automation of network monument and control, many of which have used AI techniques to some degree. However,these techniques have tended to be embedded in otherwise classical distributed computing architectures. This discussion (ahem) will consider some ideas on how - and why - we might view the overall system as a thinking thing with some empirical and engineering results form our recent work.

  • March 31
    Speaker:Richard Gold (Computer Science, Uppsala University)
    Title: Underlay Networks (slides (pdf))
    Abstract: We see the lack of adjustable indirection as being a key problem with today's Internet. Currently, there is only one way that forwarding, routing and service addressing can occur in an IPv4 network. We wish to reach into the network and make these mechanisms explicit. We propose an underlay network (a network below the network layer) as one way of acheiving these goals.

    We imagine that the underlay would have native support for building custom, bespoke networks such as: Peer-to-Peer networks, Content Delivery Networks (CDNs), Proxy networks etc. Our approach is to sit underneath the IP stack (or other network layer). Do not confuse the stack above with temporary structures and state. We can easily build a temporary network on-the-fly which can then be removed when not needed - particularly suitable for ad-hoc routing. We have already built such a system dubbed LUNAR.

    Current work is based on building custom network instances (WaveLAN, Ethernet, UDP, PPP etc.) with looking towards how translating between multiple instances in a node can take places. The objective of this is to enables clouds comprising of different network instances can be brought together on an ad-hoc basis.


April 2004

  • April 28(Room 214) (first week of term)
    Speaker:Rev Adrian Kennard(Andrews and Arnold Ltd)
    Title: Broader Broadband
    Abstract: Making the best use of multiple broadband internet connections to provide better speed and reliability.
    The talk addresses the different requirements for increased speed and reliability and the way different solutions meet these differing requirements. It is focused mostly on solutions from products and services we offer, but provides a good general overview of the issues.


May 2004

  • May 05
    Speaker:Prof. Polina Bayvel (UCL)
    Title: Optical network design - dynamic vs static wavelength routing?
    Abstract: Wavelength-routing is an elegant technique for routing and processing of information in core networks and the key problems there are in optimising wavelength assignment and minimising wavelength resources utilised (ie number of wavelengths) - a kind of graph colouring problem, solvable with good heuristics where wavelength assignment can be optimised subject to a given traffic demand. Termed WRONs (wavelength-routed optical networks) - they are relatively easy to design but may not be efficient since sub-wavelengths capacities can not be addressed and a possible solution for this is the dynamic allocation of wavelengths. One approach is optical burst switching, where electronic packats are aggregated at the network edge and then assigned to wavelengths either in a hop-by-hop basis or transparently, with or without acknowledgement of the allocated resources. Are there any operational benefits with dynamic networks where resources are allocated on a request basis? Is this dependent on traffic statistics and/or traffic self-similarity? These questions, together with some results on the design of optical burst switched networks, will be addressed. Burst aggregation, dynamic wavelegth routing and request scheduling will be discussed and it will be shown that traffic self-similarity has no impact on optical network design.

  • May 12
    Speaker:Wayne Luk (Imperial College)
    Title: Reconfigurable hardware for network applications
    Abstract: The talk describes current research on exploring the use of reconfigurable hardware for network applications. There has been rapid advances in density and capability of reconfigurable hardware such as Field-Programmable Gate Arrays. I shall provide an overview of a method for mapping high-level descriptions, captured in the Ponder language, into packet filters in reconfigurable hardware to produce internet firewalls. Other applications targeting reconfigurable hardware will also be described.

  • May 19 (slot taken)
    Speaker:Antonio Carzaniga , University of Colorado at Boulder
    Title:Content-Based Networking : Design of a Content-Driven Communication Service
    Abstract: Content-based communication is a communication service whereby the flow of messages from senders to receivers is driven by the content of the messages, rather than by explicit addresses assigned by senders and attached to the messages. Using a content-based communication service, receivers declare their interests by means of selection predicates, while senders simply publish messages. The service consists of delivering to any and all receivers each message that matches the selection predicates declared by those receivers. Content-based communication subsumes a service commonly known as publish/subscribe event notification. This talk is about the design of a content-based communication network. I will first introduce the basics of content-based networking, covering the service model and the network architecture. I will then present a routing scheme and aforwarding algorithm that realize the stated service model.

  • May 25
    Location: Room 214
    Time:4 pm
    Speaker:Jeremiah Scholl(Media Technology Division,Luleå University of Technology)
    Title:An HCI Approach to Scalability in Video Conferencing
    Abstract: The goal of providing desktop video conferencing to larger and larger groups has received a vast amount of attention from the research community since the early 90s. In general, the majority of progress has occurred in two areas, these being video compression, which reduces the size of each individual stream, and multicast delivery (IP Multicast and overlay networks), which seeks to efficiently deliver a single stream to multiple users. Despite a fair amount of progress in both areas, today it is still relatively difficult to serve groups with a few dozen or more users. The seminar will focus on the Human Computer Interaction side of the problem, and present a few ideas on how well designed user interfaces, and complimentary bandwidth sharing schemes based on human communication patterns can be used to help satisfy larger groups, independent of the codec or delivery system used. In addition, the seminar will introduce an upcoming study that we will be conducting here at UCL (with the hope of getting some feedback :)) into heterogeneous bandwidth sharing, with the goal of showing the benefits that can be provided by providing a larger share of session bandwidth to a subset of "important" group users.

  • May 26
    Speaker:Peter Kirstein
    Title: Regional Satellite Internet Access for the S. Caucasus and Central Asia The SILK Project - slides (ppt)
    Abstract: The ancient Silk Road was not only a trade route but also an all-important road for the transfer of information and knowledge between major regions of the world. The SILK Project is bringing cost effective, global Internet connectivity to nine former Soviet republics of the Caucasus and Central Asia through state-of-the-art satellite technology, thus creating a virtual Silk information highway. The aim of the SILK Project is to increase significantly the exchange of information with, and between, academic and educational institutions in these regions. It is the first time that a NATO-sponsored initiative is being managed under an EC project with significant support from Cisco and others. The network is now operational in eight countries, and we are considering the next stage of the project. The talk will describe the background of the project, the network technology, what services are provided, and both the technical and organisational governance. There are some technical novelties like the way bandwidth is shared, caching, some use of Voice/IP and experiments in IPv6. It has had significant impact including video conferences involving several Heads of State. It will also describe a number of related training activities.


June 2004

  • June 02
    Speaker:Prof. Fred Piper (Information Security Group, Royal Holloway)
    Title: Authentication
    Abstract: IIdentifying users is a central problem when using distributed systems. The limitations of using cryptography based solutions and/or the something known plus something known approach have 'always' been acknowledged. We will look at some of the problems with some proposed solutions. We might end, (or even start!), with an open discussion on National ID cards, given sufficient interest. (OHP needed)

  • June 09 (last week of term)
    Speaker:Iain Phillips (Lecturer, Computer Science, Loughborough University) and Jose Hernandez (PhD student, Computer Science, Loughborough University)
    Title: Measuring and Modelling the Internet
    Abstract:It is widely recognised that the networks community needs accurate models of the traffic behaviour Internet and other large networks. It is further recognised that such models should be based on measurements of the existing Internet. Researchers at Loughborough have been involved in Network Measurement for over 10 years, both as a research topic and providing measurement applications to BT operations.
    In this seminar we present some techniques to characterise Internet delays. The work aims to find a simple mathematical represenation, i.e. distribution parameters, for the end-to-end performance of Internet links. Our current model is based on a weighted combination of Weibull distributions. We will present the model, techniques for discovering the best combination of parameters, some theoretical backing and discuss possible applications.


July 2004

  • Tuesday July 13, 12 noon, room 203
    Speaker:Ran Atkinson (UCL CS)
    Title: Naming and Addressing in the Internet
    Abstract: This talk provides a quick survey of the existing namespaces in the Internet today. Then it suggests that there are problems with the current architecture and very briefly outlines a prospective approach that would add a new namespace to the Internet Architecture.

  • July 28
    Speaker:Ian Wakeman (Department of Informatics,University of Sussex)
    Title: "Policy management of pervasive computing services through natural language" with David Weir, Bill Keller, Julie Weeds, Tim Owen and many more (slides) Nathab project page
    Abstract: In this talk I will describe ongoing work in a multi-disciplinary project to develop a policy management architecture for pervasive computing. Our main driver is the need to provide a natural language interface, so that speech can eventually be used as the main input channel. We are therefore focussing on developing a system which is designed around the use of natural language to specify how to control the interaction and use of services. We have taken a corpus of policies from various people and used these to drive the design of a policy description logic and an associated ontology. Simultaneously, we have been working on a novel service composition architecture, based on events and late binding dependent on context. The flow of events and action requests is then controlled by the interpretation of the policy sentences. In this talk, I will describe the current state of the project, and the challenges that remain to be met.

  • Thursday July 29
    Room 214
    Speaker:Brian Levine (UMass)
    Title:Threats to Privacy from Passive Internet Traffic Monitoring
    Abstract:With widespread acceptance of the Internet as a public medium for communication and information retrieval, there has been rising concern that the personal privacy of users can be eroded by malicious persons monitoring the network. A technical solution to maintaining privacy is to provide anonymity. There have been a number of protocols proposed for anonymous network communication. We show there exist attacks based on passive traffic monitoring that degrade the anonymity of all existing protocols. We use this result to place an upper bound on how long existing protocols, including Crowds, Onion Routing, Mix-nets, and DC-Net, can maintain anonymity in the face of the attacks described. This provides an analytical measure by we can compare the efficacy of all protocols. Our analytical bounds are supported by tighter results from simulations, and we made empirical measurements of our assumptions. We found that mix-based protocols offer the best tradeoff of performance and security.

    In our most recent work, we have looked at further problems with protecting privacy: attacks to detect signatures of users and webservers that persist over days or weeks. VPNs created by ssh tunnels or secure wireless connections (e.g., WEP) as implemented are not sufficient to block these signatures, even though they provide more protection than SSL-based connections that have been looked at previously for the same problem. We designed an attack and evaluated it with real Internet measurements: allowed a training period, we found an attacker could guess which exact web site (in the training set) was visited by a user through an encrypted link almost 40% of the time; 70% of the time the correct answer was in the attacker's top five guesses.


August 2004

  • August 11
    Speaker:Venus Shum (UCL, EE)
    Title: A cluster-based approach for data handling in self-organising sensor networks. slides (ppt)
    Abstract: With the advances in wireless technology coupled with intelligent devices, distributed and self-organising sensor networks have gained increasing importance for environmental monitoring. In the case of the SECOAS oceanographic monitoring project, a sensor network provides a viable and low cost alternative to the traditional expensive and large sensor packages used in oceanography.

    One of the major challenges of SECOAS is the development of an appropriate data handling technique, which is required to preserve significant phenomena within the data for the oceanographers, whilst remaining power efficient. We present a cluster-based approach to the problem of spatial sensing and data fusion. We place emphasis on the creation of clusters of nodes with similar spatial attributes by the use of distributed algorithms mimicking some biological entities, for example quorum sensing. These algorithms are supported by a light-weight operating system, kOS. Additionally, the self-organising behaviours of the clustering algorithm will also be discussed in the presentation.

  • [Top]

September 2004

  • September 22
    Speaker: Mark Handley/Saleem Bhatti
    Title: Plans for HEN (Heterogeneous Experimental Network)
    Abstract: HEN is intended to be a shared facility for running networing experiments. This should remove the need for half of us to have mini-clusters under our desks - which won't be possible in the new building. The hope is that it should become much easier to do experiments without spending ages acquiring hardware, setting up an OS, ensuring security, etc.

    We'll discuss our plans for HEN, what we've done so far, next steps, and longer term ideas. The reason for discussing this now is to sanity check that what we think we need is what YOU think we need. So come along and brainstorm about what you might want from experimental networking infrastructure, and how you might help out.

  • September 29 Nets Group Meeting
    Location : G205 in Gower Street (number 66-72, second floor) NOT THE PEARSON BUILDING
    Time : 2pm


October 2004

  • October 6
    Speaker:Rae Harbird
    Title:Adaptive Resource Discovery for Ubiquitous Computing paper (pdf),slides (ppt)
    Abstract: The terms pervasive and ubiquitous computing are used to describe a smart space populated by hundreds of intelligent devices that are embedded in their surroundings. Characteristically, ubiquitous computing devices must blend into the background, unobtrusively collaborating to provide value-added services for users. Services are thus essential to the success of this technology and, as a result, both service discovery and service management will play a vital role in generating the revenue stream that is a prerequisite for sustainable ubiquitous deployment. On the one hand, the services provided should be evident by their richness and variety and on the other, the complexity inherent in the environment must be hidden from users. In this paper, we describe RUBI, a resource discovery framework for ubiquitous computing. RUBI represents anovel approach to resource discovery, because the primacy of the need for adaptive autonomic behaviour is established within its design.

  • October 14(THURSDAY) Room 214 3pm until 5pm
    Speaker:Ken Carlberg(Johns Hopkins University / SAIC)
    Title: Emergency Communications (slides ppt)
    Abstract: The events of September 11, 2001 have raised the awareness and need to communicate as needed and where possible during emergencies and go beyond television and radio as sole sources of information. This seminar uses several perspectives in covering the topic of emergency communications. Subjects such as the relative importance of certain communication (e.g., who has priority and how is that priviledge granted), examples of existing prioritized emergency communication systems, and a discussion of the Internet in the context of emergencies are presented.

  • October 20
    Speaker:Ian Brown (UCL, CS)
    Title: Who's watching you on the Internet(slides ppt)
    Abstract: The last two decades were a time of momentous change for government eavesdroppers. The UK communications market shifted from a large government-owned monopoly provider, to competing large private telcos, to a fragmented market containing hundreds of firms. The technologies in use changed from mainly analogue systems through to digital networks running SS7 and on to IP systems. Finally, the privacy rights in the European Convention on Human Rights have had an increasingly strong application through the courts.
    These changes have shaken up the previously cosy world of communications intercepts. Whereas the Post Office (and the 'non-existant' GCHQ) would quietly wiretap communications on the say-so of senior police and intelligence officers, a range of new legislation has had to be put in place to maintain this "intercept capability" for intelligence and law enforcement agencies given modern networks and human rights legislation.
    This seminar will discuss this legislation and the effect it has had on ISPs' and phone companies' networks as well as the privacy of UK citizens.
    Speaker's Bio.: Ian Brown is an honorary research fellow at UCL, from where he received a PhD in the field of communications security. He has spoken and written extensively on communications and healthcare privacy, copyright and e-voting. Dr Brown is advising the US government on the security of their next-generation emergency communications systems, and is the co-author of a recent Kluwer book on this subject. He has consulted for other large organisations such as the BBC, JP Morgan and Credit Suisse. Brown is also a trustee of Privacy International and on the advisory boards of Creative Commons UK and the Foundation for Information Policy Research. He is a Fellow of the Royal Society of the Arts and a member of the Royal Institute of International Affairs, the International Institute for Strategic Studies and the Association for Computing Machinery.

  • October 27 Nets Group Meeting
    Location : G205 in Gower Street (number 66-72, second floor) NOT THE PEARSON BUILDING
    Time : 2pm


November 2004

  • November 03
    Speaker:Bob Briscoe, BT
    Title: Shared control of networks based on re-feedback slides (pdf)
    Abstract: Accurately characterising paths is the foundation of both resource sharing and routing in packet networks. Re-feedback is a simple realignment of metrics so that data headers characterise their downstream, rather than upstream path. Then the choice between network and end-point control can be made at run-time rather than prejudged as part of the network architecture. The approach is to use mechanism design - a branch of game theory. Incentives are arranged to ensure honesty and responsibility will be the dominant selfish strategies, even for brief flows, enabling natural solutions to problems in congestion control & policing, denial of service and routing.

  • November 10 : Room 214
    Speaker:Arjuna Sathiaseelan (Kings College, Dept. of Computer Science)
    Title:EPDN: Explicit Packet Drop Notification and its uses. slides (ppt)
    Abstract: The Internet is experiencing an exponential growth in users and network traffic. As the Internet grows larger and larger, the performance of the network is subjected to severe performance degrading issues such as congestion in the network, corruption of packets and reordering of packets. Reordering or Corruption of packets decreases the TCP performance of a network, mainly because it leads to overestimation of the congestion of the network. Thus it is imperative to propose a mechanism that allows the TCP sender to know the exact cause of the out of order data. We propose a mechanism called the Explicit Packet Drop Notification (EPDN) that allows the gateway to inform the TCP sender about dropped packets. We also propose two new extensions to TCP called the Reorder Notifying TCP (RN-TCP) for wired networks and Robust TCP (TCP-R) for error prone long delay networks, that use the EPDN mechanism to improve the performance when packets get reordered or corrupted in the network.

  • November 16 : Room G22
    Time : 2pm
    Speaker:Ian Brown (Dept. of Computer Science)
    Title:The politics of control and the Internet slides (ppt)
    Abstract: As the Internet became popular in the mid-Nineties, it was often thought of as a Wild West environment where few laws applied. Censorship would be routed around; privacy would be protected with unbreakable cryptographic codes; nobody would know if you were a dog.

    Ten years on, it is obvious that governments and corporations are very keen indeed to apply real-world controls to cyberspace. Attempts to restrict the use of encryption have largely failed, but other methods of surveillance are being developed and deployed. Patent and copyright laws have been "updated" for the Internet age and wielded gleefully against open source software and file sharers. Countries such as Saudi Arabia and China block their citizens' access to large portions of the Web.

    Can the tension between online freedom and control be resolved, or will there be an ongoing battle between law enforcement agencies, intellectual property owners and libertarian programmers?

  • November 17
    Room 203
    2.30pm until 3.30pm
    Speaker:Judith Johnson (Thales Group)
    Title:Policy Based Management for Naval Networks and High Assurance Environments (slides (pdf))
    Abstract: PBM is the answer! Or is it ? ....

    This seminar will introduce the major issues associated with securing and managing networks in high assurance environments, using naval coalition networks as an example. Policy based management, a popular and widely applicable delegation technique, is a potential solution to some of the management issues associated with dynamic networks (and systems) of the future, even in high assurance environments. However, there are limitations and constraints to its use.

    The seminar will present the Thales Research & Technology (UK) (TRT) approach to using PBM in such environments. A number of existing demonstration systems will be described and an indication will be given of some ongoing research at TRT.

  • November 22 - Monday
    Location : Cruciform LT2 (please bring your UCL ID cards)
    Time : 3pm to 5pm
    Speaker:Christophe Diot, Intel Research UK
    Title: ad-hoc google - Haggle
    Abstract:The Internet is built around the assumption of contemporaneous end-to-end connectivity. This is at odds with what typically happens in mobile networking, where connectivity is intermittent. We propose *opportunistic networking*, a communication model which reflects the reality faced by the mobile user. We describe the challenges that this approach entails and provide evidence that it is feasible with today's technology. We present Haggle, our user-centric architecture and implementation of opportunistic networking.
    A paper about Haggle submitted to HotNets III

  • November 24Nets Group Meeting
    Location : G205 in Gower Street (number 66-72, second floor) NOT THE PEARSON BUILDING
    Time : 2pm
    Speaker:Saleem Bhatti
    Title: Information about several projects (46paQ, MASTS. EASLEA and UKLight) (slides (pdf))

  • November 30DCNDS Seminar
    Location : G22
    Time : 2pm
    Speaker:John Souter, CEO of LINX
    Title: Monitoring at LINX (slides (pdf))
    Abstract: Explaining the overall network monitoring issues at LINX with a view to what the LINX are aiming for in the future (including some work already in progress).

  • [Top]

December 2004

  • December 01
    Speaker:Yangcheng Huang (UCL CS)
    Title: Scalable Signaling Underlay for Overlay Networks (paper (pdf)) (slides (ppt))
    Abstract: This paper presents the design of a scalable decentralized signaling underlay infrastructure, which features with DHT based management information storage and query-based state lookup mechanism. The signaling underlay is aimed to apply a decentralized peer-to-peer style searching and discovering engine into management and control plane of the overlay network, including grid networks and p2p applications, to facilitate deployment of QoS service.

  • December 08
    Speaker:Bob Briscoe, BT
    Title: Overview of BT's Networks Research Centre (slides (pdf))
    Abstract: A romp through BT projects of relevance to the NRG:
    • Overall themes:
      • Motivation
      • The science of computational networking
    • Current projects:
      • Open Spectrum potential (Collaborative phased arrays, etc)
      • Motivation Management in Peer to Peer Services
      • Privacy with pervasive computing
      • 2020 Communications Architecture - sub-projects on:
        • The science of computational networking
        • Shared control; shared value
        • Routing, naming and addressing
      • IP e2e QoS
      • Deep packet inspection (applicability)
      • BT's 21st Century Network programme
    • Some historic projects (watchcast (Global event messaging), M3I (Market Managed Multiservice Internet), mWare (multicast endpoint middleware), InternetMart (when we invented Web e-commerce but forgot to patent it)
    • Likely future directions

  • December 15 (last week of term)
    Speaker:Damon Wischik (joint work with Gaurav Raina (Cambridge))
    Title:Queueing theory and TCP congestion control
    Abstract: In large multiplexers with many TCP flows, the aggregate traffic flow behaves predictably; this is a basis for the fluid model of Misra, Gong and Towsley and for a growing literature on fluid models of congestion control. In this paper we argue that different fluid models arise from different buffer-sizing regimes. We consider the large buffer regime (buffer size is bandwidth-delay product), an intermediate regime (divide the large buffer size by the square root of the number of flows), and the small buffer regime (buffer size does not depend on number of flows). Our arguments use various techniques from queueing theory.

    We study the behaviour of these fluid models (on a single bottleneck link, for a collection of identical long-lived flows). For what parameter regimes is the fluid model stable, and when it is unstable what is the size of oscillations and the impact on goodput? Our analysis is based on an extension of the Poincar-Linstedt method to delay-differential equations.

    We find that large buffers with drop-tail have much the same performance as intermediate buffers with either drop-tail or AQM; that large buffers with RED are better at least for window sizes less than 20 packets; and that small buffers with either drop-tail or AQM are best over a wide range of window sizes, though the buffer size must be chosen carefully. This suggests that buffer sizes should be much much smaller than is currently recommended.