Seminars

For many years the UCL SSE group has run a successful series of seminars, showcasing both mature and developing research from our own researchers and visiting speakers.

The seminars are currently scheduled regularly at 1pm on Mondays during term-time.  To receive email notifications of upcoming seminars, including titles and abstracts for the talks, please send an email to sse-seminars-request@cs.ucl.ac.uk.

October 2007



• August 14
  Speaker: Mohamed Ahmed
  Title: A policy-based management framework for networked embedded systems: The RUNES approach
  Abstract: Policy management and enforcement mechanisms are required to ensure that the devices and systems that make up a pervasive environment function as intended. However, though a number of proposals exist to provide mechanisms which function in this capacity, they make the following tacit assumptions: (i) We can always rely on the existence of some policy management framework in the environment, (ii) there is always some prior knowledge that we can use to write application or component rules, otherwise we may use real-time event handling mechanisms to constrain the behaviour of applications or components.

  We believe that in reality these assumption are too strong. For pervasive computing environments such as the scenarios envisioned in the RUNES project, we have decentralised computing environments with multitudes of devices with different capabilities and no clear organisational hierarchy. Further the rapid adoption of component based systems to provide applications and support mobility and runtime reconfigurability means that assumptions of complete prior knowledge cannot be supported.

  To address these limitations, this work aims to decouple the policy management and enforcement tasks in order to take advantage of available specialisations and cater for a wider variety of systems.

  We present a mechanism that support an ontology based policy definition and provide the capacity to statically verify the satisfiability of policies. Once policies are declared viable, rather than passively monitoring whether the constraints set are respected, our mechanism binds the constraints in the policy rules to the components/primitives they refer to and these components may then be distributed as stand-alone applications with built in policy enforcement.
  
  

November 2007



• November 19
  Speaker: Arun Mukhija
  Title: The Dino Approach to Dynamic and Adaptive Service Composition
  
  Abstract: In this talk, I will give an overview of our work on the Dino project. The aim of the Dino project is to provide enabling technologies for dynamic and adaptive composition of autonomous services. This includes providing support for all stages of service composition: from the specification of service requirements and capabilities, to runtime service discovery, selection, binding, delivery, monitoring and adaptation. Our work in Dino is particularly targeted towards open dynamic environments, where new services may appear and old ones may disappear unpredictably, and where runtime changes in the availability of resources or service requirements are a norm. In this talk, I will describe the overall approach of Dino, with a special focus on the QoS-aware and self-adaptive service composition features of Dino. I will also outline the work we plan to do in the coming months.
  
  
• November 22, 3pm - 4pm, Roberts Building, Malet Place, WC1E 6BT, Room 106
  
  Speaker: Miron Livny, Dept. of Computer Sciences, University of Wisconsin
  
  Title: Submitting locally and running globally – The GLOW and OSG Experience
  
  Abstract: The Grid Laboratory of Wisconsin (GLOW) is a NSF funded, distributed facility at the University of Wisconsin – Madison campus. It is part of the newly formed Center for High Throughput Computing (CHTC) and consists of more than 1800 processing cores and 100 TB of storage located at six different sites. Since its inception in the winter of 04, it has been serving a broad range of disciplines ranging from Biotechnology and Computer Sciences to Medical-Physics and Economics. Each of the GLOW sites is configured as an autonomous locally managed Condor pool that can operate independently when disconnected from the other sites. Under normal conditions, the six pools act like a single Condor system that is coordinated via a highly-available campus-wide matchmaking service. On-campus and off-campus users interact with GLOW through job-managers located on their desktop computers.

  The Open Science Grid (OSG) is a DOE and NSF funded US national distributed computing facility that supports scientific computing via an open collaboration of researchers, software developers and computing, storage and network providers. The OSG Consortium is building and operating the OSG facility, bringing resources and researchers from universities and national laboratories together and cooperating with other national and international infrastructures to give scientists access to shared resources world-wide. The particular characteristics of the OSG are to: Provide guaranteed and opportunistic access to shared resources; operate a heterogeneous environment both in services available at any site and for any Virtual Organization, and multiple implementations behind common interfaces; Support multiple software releases at any one time; Interface to campus and regional grids; Federate with other national and international grids.

  In the talk, we will present the principals that guided us for more than two decades in developing our distributed computing technologies and high throughput computing software tools and will provide an overview of these two high throughput computing facilities. Capabilities to “elevate” local GLOW jobs to the national OSG infrastructure will be discussed. These capabilities follow our long standing “bottom-up” approach to the construction and operation of large scale distributed computing infrastructure that maximize reachable capacity while preserving local access, environment and autonomy.
  
  

• November 26
  Speaker: Bruno Wassermann
  Title: Increasing the Availability of Large-Scale Service Compositions Through Autonomic Failure Diagnosis
  
  Abstract: In this 45 minute talk I will present what I have dared to submit for my first year viva. I will present the problem of ensuring reliability and availability in large-scale service compositions built as Web service orchestrations in BPEL, demonstrate that these issues have not been solved adequately so far, explain why this is so and why it's important we find a solution.

  We will look at the various parts comprising my proposed contribution whose aim it is to increase the cooperation among the components involved in service compositions when handling failures and reduce the problem space users are confronted with when diagnosing failures.

  Finally, I will outline some initial thoughts on how to evaluate the benefits/effects of this approach and characterise several quality attributes of interest.
  
  

December 2007



• December 10
  Speaker: Franco Raimondi
  Title: PDVer, a tool to verify PDDL planning domains
  
  Abstract: We present PDVer, a tool for the verification of planning domains written in the Planning Domain Definition Language. The tool is an Eclipse plug-in and it accepts a PDDL domain in input. Users can enter requirements (flight rules) expressed as LTL formulae. The tool applies extended structural coverage metrics to each requirement, to derive a set of trap formulae characterizing the different ways in which that requirement can be met or failed. The trap formulae are expressed as additional planning goals for the original domain and analyzed using the planner. In this way, PDVer enables the automatic, coverage based verification of a large class of requirements on PDDL models. Differently from other approaches, our tool does not require a model checker to perform verification, but it reduces the problem of verification to an extended planning problem. (This is a joint work with Guillaume Brat and Charles Pecheur at NASA Ames)
  
  

Upcoming seminars

January 2007

• Wednesday, January 30th, 1-2pm, rm 508, Roberts Building
  Speaker: Alberto R. Schiesari
  Title: Legacy Technologies in Corporate Environment
  Abstract: Even today, most of large corporations (in both developing and developed countries) still rely on 'legacy' technologies such as mainframes for daily operations. In this talk, I will present a review of the characteristics of the mainframes and related technologies, the suppliers, the profile of the users, and the roles that these technologies play, including software development and maintenance. Also, I will discuss some trends in corporate environment concerning the use of legacy systems.