- October 6
Speaker:Wolfgang Emmerich
Title: Publishing Software Engineering Research
Abstract: Conducting research without reporting about its results is like not doing the research in the first place! Publications are one of the key performance indicators for the success of both individual researchers and research groups. Funding available for Higher Education Institutes as well as individual researcher's salaries are directly correlated to publication track record. In this talk, we discuss how to prepare research results so that they become publishable. We discuss the importance and standing of different journals and conferences for software engineering research. We look at how decisions or acceptance are being made and discuss the criteria that are being considered for that condition in an attempt to improve our publication track record, individually and as a group.
- October 13
Speaker:Ben Butchart
Title: The Open Grid Services Architecture
Abstract: The Open Grid Services Architecture (OGSA) is a new specification for developing Grid applications using web services technologies such as SDL and SOAP. Our experience re-engineering an existing scientific application as a set of loosely coupled OGSI Web Services has given us some insight into the difficulties of modelling real applications in this environment. One such difficultly is the need configure and control the verall computational process to ensure efficient use of available resources and to provide fail over, load balancing and recovery. We are evaluating the Business Process Execution Language (BPEL), a new specification for choreography and orchestration of Web Services, to determine whether this standard could work for OGSI-enabled Grid applications. We discuss the shortcomings of existing BPEL implementations and of the language itself in meeting the special demands of Grid applications. We explain how these shortcomings can be overcome with a little imagination. We also compare BPEL to other workflow languages that could be used for modelling service composition of Grid applications.
- October 20
Speaker:Mirco Musolesi
Title: Developing a Middleware for Ambient Intelligence
Abstract: Ubiquitous computing is now a reality and various classes of applications are available. It is possible to classify these applications according to the types of targeted end-users. For instance, Ambient intelligence systems are targeted to a large public and particularly oriented towards infotainment applications. People have a perception of the systems as intelligent, since these adapt automatically to the types of devices that they are using (for example fixed, mobile or embedded into a car) and to their preferences. In this seminar we will discuss some aspects of the development of the
Ozone WSAMI environment, a Web services based middleware solution for Ambient Intelligence, focusing in particular on the issues related to
the software components especially designed for resource-constrained terminals. - November 10
Speaker:Andrea Polini
Title: Testing CBSE, new issues and new possibility
Abstract: Component based development is the emerging paradigm in software production, but several testing issues are still open. In fact the "old" testing methodologies are hardly applicable due to a general lack of information (particularly strong with COTS) concerning the component structure. Moreover, tests performed by the developer cannot give enough guarantee to the customer that can deploy the component in a different environment. As a consequence specification customer testing has been advocated. To this purpose, we have developed the CDT framework for Component Deployment Testing. CDT provides the customer with both a technique to early specify a deployment test suite and an environment for running and reusing the specified tests on any component implementation. At the same time new testing applications seem quite promising. In particular, recently a possible test based approach for the early evaluation of distributed software applications has given promising results.
- November 17
Speaker: Costantinos Costantinides
Title: Providing multidimensional decomposition in object-oriented analysis and design
Abstract: Despite the success of OOP towards achieving separation of concerns, certain properties cannot be cleanly modularized, imposing the symptoms of code tangling and code scattering. Aspect-Oriented Programing (AOP) technologies (like AspectJ) provide linguistic mechanisms in order to explicitly capture crosscutting concerns (aspects) during implementation and special compilers (weavers) in order to combine them with components.In order to produce a clean (tangled-free) implementation and achieve the maximum benefits of advanced separation of concerns, analysis and design artifacts themselves must in turn explicitly address crosscutting concerns. We therefore need to provide the means to identify and model crosscutting concerns from the early stages of software development and make them traceable. To this end, Aspect-Oriented Software Development (AOSD) has extended AOP to include aspect-oriented activities throughout software development.
The motivation behind this work is to show how AOSD can be utilized (and prove successful) in every phase of development, so that the explicit capture of crosscutting concerns in code should be the natural consequence of good and clean modularity and not the result of a corrective measure due to a tangled implementation. In this talk I will discuss a case study in order to investigate how crosscutting concerns can be identified, propagate and can be handled throughout development. The main theme of this talk is to show how to adapt established analysis and design techniques for object-oriented systems to an aspect-oriented context.
- December 1
Speakers: Clovis Chapman
Title: Resource sharing in overlay grids
Abstract: In the context of the eminerals e-science project, the need for preferential access to a number of physical resources made available by the different institutions involved in the project has led to the establishment of the eminerals 'mini-grid': a computational grid built from resources on the UK level 2 grid for the sole use of members of the eminerals team. Our experience in establishing this overlay grid has brought us to investigate how resource allocation and scheduling policies can be defined by resource owners in the presence of overlay grids and how these policies can be enforced whilst respecting site autonomy. We will discuss some of the shortcomings of current grid technology used in the project, define some of the principles and trends in e-science grid computing and outline some aspects of our investigation.
- December 8
Speakers: Usman Arshad
Title: Mobile Healthcare (Results of TCS project between UCL and Capula Elan)
Abstract: Small lightweight mobile devices are increasingly seen as the ideal platform to revolutionise the use of IT in healthcare by providing compute power at the 'point of care'. Such devices are suited to the physically mobile clinical environment and by using wireless communications to form part of an enterprise system, can become extremely powerful tools that are both informational and interactive. The result of this TCS program is m-oasis - an 'electronic patient record' (EPR) application for PDAs that uses wireless communications to interface to the enterprise. m-oasis gathers context information about network availability and adapts between online and offline modes of operation, and subsequently deals with any conflicts that may occur as the result of an inconsistent view of data.
- December 15
Speaker: Genaina Rodriguez
Title: Reliability Support for the Model Driven Architecture
Abstract: Reliability is an important concern for software dependability. Quantifying dependability in terms of reliability can be carried out by measuring the continuous delivery of a correct service or, equivalently, the mean time to failure. Our intended contribution is to provide a means to support reliability design following the principles of the Model Driven Architecture(MDA). By doing this, we hope to contribute to the task of consistently addressing dependability concerns from the early to the late stages of software engineering. Additionally, we believe MDA can be a suitable framework to realize the assessment of those concerns and therefore, semantically integrate analysis and design models into one environment. In this presentation we elaborate our approach on how the provision of reliability can be realized through a standard model-driven architecture approach. - January 19
Speaker: George Roussos (Birbeck)
Title: Service Provision in a Ubicomp World: The Planet Lab Way
Abstract: Ubiquitous computing's aim is to activate objects with computational and communications capabilities. Recent experience has indicated that ubicomp-augmented artifacts naturally create the need for geographically distributed network services which are developed as overlays on the Internet. Thus, the boundary of application processing and packet forwarding is blurring and a new service-oriented network architecture (SOA) is emerging. Planet Lab is such an architecture and a testbed build around four principles: services should be able to run continuously and access a slice of the SOA bus resources, control over resources should be distributed, management services should be unbundled and run in their own slices, and APIs should be designed to promote application development. In this talk, I will first present two case studies of ubicomp systems which benefit significantly form the emerging SOA, namely the Electronic Product Code EPCglobal trading network for consumer goods and Smart Dust networks for environmental monitoring. Then, I will discuss the Planet Lab architecture from a node perspective with emphasis on the current breed of Planet Lab virtual machines and their implementation.
- January 28
Speaker: Carina Alves
Title: Managing Mismatches in COTS-Based Development
Abstract: The prospect of reducing the time, cost and risk of system development has increased the interest in developing systems from COTS Commercial-Off-The-Shelf) software products. The selection of COTS is a complex activity that requires careful assessment of how stakeholder requirements are satisfied by products. Due to the generic nature of COTS, it is possible that mismatches may occur between what is desired by the stakeholders and what it is possible to achieve. To complicate matters, the available information about the software comes from incomplete and contradictory sources. This gives rise to risks that may prevent the successful deployment of the system. We have developed the TAOS (Tradeoff Analysis forcOts-based Systems) method to guide COTS decision-making. The novelty of the approach is twofold. First, it supports the explicit treatment of tradeoffs through the process of mismatch analysis. Second, it provides an exploratory strategy to discover and handle risks while reasoning about uncertainty. - February 2
Speaker: Rocco Moretti (University of Bologna)
Title: Towards a Model Driven Method for Ranking and Querying Grid Services
Abstract: n the forthcoming Grid architecture, core Grid services providing computing, storage, and network capabilities will be the building blocks for deploying complex services based on electronic contracts meant as a machine-readable representation of obligations useful for computer-supported contracting and fulfillment. In this context, refining mechanisms for ranking, and querying such services is meaningful. The ranking process consists in defining criteria enabling to select one service among those satisfying a set of constraints, while the querying process concerns techniques for obtaining service statistics useful for monitoring and evaluating services. The aim of this presentation is to propose a method supporting this activity by integrating three related concepts: a solid software measurement theory, the Logic Scoring of Preferences (LSP) method for evaluating, comparing, and selecting software systems, and the metamodeling hierarchy suggested by the Model Driven Architecture (MDA).
- February 9
Speaker: Martin Hailey
Title: Still worrying about concerns
Abstract: All software is mostly rubbish. This is good because it's a situation that needs improving and therefore we all have jobs. Aspect-oriented software development (AOSD) technologies aim to provide a means of improvement by allowing the modularization of what we refer to as crosscutting "aspects" or "concerns" of system development. The intention is that this will then facilitate more efficient creation, evolution, understanding, integration, etc of a system. AOSD concepts can be applied at any stage of the software lifecycle, but the nature of the technology does not lend itself wholly to established techniques and practices.Our work looks at how AOSD techniques can be effectively and correctly used in conjunction with the UML. This is a significant problem, or at least one that needs procrastinating over for a couple of years, so come along and witness my delusions of adequacy as I proffer half a solution to a half-baked problem.
- February 23
Speaker: Group Discussion
Title: Future of the SSE Seminars - March 1
Speaker: Andrew Dingwall-Smith
Title: Run-Time Requirements Monitoring for Goal-Oriented Requirements Specifications
Abstract: Software systems are built based on assumptions about the environment in which they will operate. Changes in the environment can therefore result in failures of the system which cannot easily be anticipated.Runtime requirements monitoring confirms, as a system operates, that it is satisfying the requirements specified for it. Requirements monitoring as part of the normal operation of a system allows unanticipated failures to be identified, and rectified through system evolution.
Our work is in run-time monitoring of goal-oriented requirements specifications, using temporal logic. This is achieved by instrumenting the system, using AspectJ, to emit events. These events are used to detect requirement violations. The instrumentation aspects are automatically generated using a mapping from the requirements specification to the system design.
Another concern is monitoring soft goals; goals which do not have formal criteria for satisfaction. The problem here is determining metrics which can be monitored so that a determination of satisfaction can be made.
- March 8
Speaker: Cecilia Mascolo, Stephen Hailes and Karen Lawson
Title: RUNES - Reconfigurable Ubiquitous Network Embedded Systems
Abstract: The talk will describe the new European Project RUNES which will (hopefully) start in the summer. After a quick description of how project funding works in EU Framework 6 and of the possible instruments, the talk will illustrate the main focus of the RUNES project and the challenges that the Embedded Systems environment poses to middleware, network, HCI research.
- March 15
Speaker: Ofer Margoninski
Title: A middleware for integrating heterogenous models in systems biology
Abstract: We review the requirements for computer modeling of heterogenous biological systems, across multiple aspects and scales. We note the inherent organization in biological systems, on the one hand, and the great heterogeneity and diversity of those systems, as well as of the modelling approaches used to model them, on the other hand. We examine the way in which models are authored and maintained, and conclude that there is a need for a modeling approach that would support modularity, componentization, the integration of heterogenous models created in different schemes and with different tools, and the representation of modeling assumptions and constraints. We propose to build a middleware architecture, probably OGSA/GRID based, that would support these needs.
- March 22
Speaker: Peter Saffrey
Title: Systems Biology continued - biological meta-modelling and the link with experimental results
Abstract: Last week, Ofer introduced the challenge of systems biology and discussed the componentisation of biological models. In this talk we will discuss some of the architecture and meta-data requirements necessary to facilitate this task.In particular we will present the latest version of our biological meta-model, which provides an overview of the salient features of biological models. This meta-model will then be developed with reference to some specific issues. Particular focus will be made on work to address the relationship between biological models and their experimental basis.
We will also discuss other interesting research topics arising from this work such as the need to address assumptions used to derive a biological model and how these assumptions allow us to check biological models for consistency.
- March 29
Speaker: Carine Courbis
Title: Towards an Aspect Weaving BPEL Engine
Abstract: In this seminar, we will present how we are building a highly configurable and extensible BPEL engine. First, we will briefly present the different technologies or mechanisms involved: the visitor design pattern, AOSD (Aspect-Oriented Software Development), SOA (Service-Oriented Architecture), and BPEL (Business Process Execution Language). Then, we will explain how the visitor design pattern and dynamic aspects can be used to obtain this BPEL engine. Using these two techniques, the core of this infrastructural software can be customized to meet new requirements and add features such as debugging, execution monitoring, or constraint checking. Additionally, it can easily be extended to cope with customer-specific BPEL extensions. We propose the use of dynamic aspects not only on the engine itself but also on the workflow in order to tackle the problems of Web Service hot deployment and hot fixes to long running processes. In this way, composing a Web Service "on-the-fly" means weaving its choreography interface into the workflow. Finally, we will conclude and give a brief glimpse of the AOSD conference that took place last week in Lancaster. - April 19
Speaker: Prof. David S. Rosenblum
Title: Some Open Problems in Publish/Subscribe Networking
Abstract: Publish/subscribe networking (sometimes also called "content-based routing") is an attempt to scale the publish/subscribe communication style to applications deployed over wide-area networks with huge numbers of publishers and/or subscribers. In keeping with the spirit of our seminars, this talk will focus primarily on unsolved problems in publish/subscribe networking. I will briefly review some of the key results from previous work in the Siena research project and at PreCache (a Sony-funded spinout), and then I'll describe a number of open problems ranging from the deeply technical to the ridiculously pragmatic. - April 26
Speaker: Stefanos Zachariadis
Title: Logical Mobility for Mobile Self-Organisation
Abstract: There is growing interest in both research and practice of self-organising systems. These systems re-organise in response to new or changing requirements. The need for self-organisation is often found in mobile applications; they are typically hosted in resource-constrained environments and may have to dynamically reorganise in response to changes of user needs, to heterogeneity and connectivity challenges as well as to changes in the execution context and physical environment. We argue that physically mobile applications benefit from the use of self-organisation primitives. We show that logical mobility primitives and a component-based approach assist in building self-organising mobile systems. We present SATIN (System Adaptation Targeting Integrated Networks), a lightweight component model, which represents a mobile system as a set of interoperable local components. The system supports reconfiguration, by offering migration services. - May 10
Speaker: James Skene
Title: Precise Service Level Agreements
Abstract: Service Level Agreements (SLAs) are part of a contract between the provider and client of some service, and describe the performance properties that the service is required to exhibit. As part of the TAPAS IST project, Davide Lamanna, Wolfgang Emmerich and I have developed a language for SLAs, appropriate for various kinds of internet services, called 'SLAng'. The advantages of using a predefined language to write SLAs are to ensure machine readability of SLAs, to make it easier to write good SLAs by reusing the definitions of the terms of the language, and to ensure that SLAs are defined in a manner that permits useful analysis of their terms. In my talk I will describe the approach taken to defining the meaning of SLAng, with the objective of making this reusable definition as precise, understandable and useful as possible. - May 19
Speaker: Rami Bahsoon
Title: Applying ArchOptions to Value the Payoff of Refactoring
Abstract: As software is enhanced, modified, or adapted to new requirements, the
software becomes more complex and drifts away from its original design. To reduce complexity, there is a need for techniques that incrementally
improves the internal software quality. The research domain that addresses this problem is referred to as restructuring, or in the case of object-oriented and agile development, as refactoring. Restructuring and refactoring are used to improve the quality of the software such as
extensibility, modularity, reusability, complexity, and efficiency. In
refactoring, the key idea is to redistribute classes, variables, and methods across the class hierarchy in order to facilitate future adaptations and extensions. This in turn will result in a modified structure (compared to the original) with different qualitative measures and value potentials.
Little has been done on understanding the economics of refactoring: when is it cost-effective to invest in a refactoring exercise? How can we value the payoff due to refactoring, prior to investing in such an exercise? How can we reason about this payoff in connection with changes in the structure and at correspondingly higher level of abstractions than code? To answer these questions, we build on ArchOptions to value the payoff of refactoring. ArchOptions is a real-options based model that we have proposed and published in the Fifth Economics-Driven Software Engineering workshop (EDSER 5) to value the flexibility of software architectures in response to likely future changes in requirements. In the EDSER 6 paper, we build on ArchOptions to devise an options-based model to value the architectural flexibility that results from a refactoring exercise. This value assists in understanding the payoff of investing in refactoring: that is, if the refactored system results in an architecture that is more flexible, such that the expected added value (in the form of options) due to the enhanced flexibility outweighs the cost of investing in this exercise, then refactoring is said to payoff. We apply our model to a refactoring case study from the literature. - June 7
Speaker: Mirco Musolesi
Title: Designing and testing context and social-aware adaptive systems for ad hoc environments
Abstract: Mobile devices are usually carried by people. Starting from this simple assumption, it is possible to study ad hoc networks considering them as social networks.
In the recent years, social networks (and complex networks in general) have been investigated in different areas such as sociology, mathematics and physics with interesting and, in a sense, surprising results.
We believe that it is possible to design systems that take in consideration the structure of the social networks that are behind the possible different deployment scenarios.
In this seminar, we will discuss the challenges in designing and testing adaptive mobile systems (and, in particular, message oriented middleware systems) that exploit the intrinsic characteristics and properties of ad hoc networks. We will present some results including a brief description of CAR, a Context-aware Adaptive Routing protocol, and our current work on the design of a realistic mobility model founded on social networks theory for the simulation and testing of context and social-aware systems and architectures.
- June 14
Speaker: Emmanuel Letier
Title: Reasoning about Partial Goal Satisfaction for Requirements and Design Engineering
Abstract: Exploring alternative options is at the heart of the requirements and design processes. Different alternatives contribute to different degrees
of achievement of non-functional goals about system safety, security, performance, usability, etc. Such goals in general cannot be satisfied
in an absolute, clear-cut sense. Various qualitative and quantitative frameworks have been proposed to support the assessment of alternatives
for design decision making; in general they lead to limited conclusions due to lack of accuracy and measurability of goal formulations and of
impact propagation rules along goal contribution links.
The talk presents techniques for specifying partial degrees of goal satisfaction and for quantifying the impact of alternative system
designs on the degree of goal satisfaction. The approach consists in enriching goal refinement models with a probabilistic layer for
reasoning about partial satisfaction. Within such models, non-functional goals are specified in a precise, probabilistic way; their specification
is interpreted in terms of application-specific measures; impact of alternative goal refinements is evaluated in terms of refinement
equations over random variables involved in the system's functional goals. A systematic method is presented for guiding the elaboration of
such models. The latter can then be used to assess the impact of alternative decisions on the degree of goal satisfaction or to derive
quantitative, fine-grained requirements on the software to achieve the higher-level goals.
- Monday 14th June marks the end of the SSE seminar sessions for this academic year. Seminars will resume again in October 2004.
October 2003
November 2003
December 2003
January 2004
February 2004
March 2004
April 2004
May 2004
June 2004
This page was last modified on 18 Oct 2013.
+44 (0)20 7679 7214 - Copyright © 1999-2013 UCL