We will further discuss architectural issues for putting repair systems into practice, and present examples of repair in the context of UML-based software development, including the repair of EJB deployment descriptors following cross-notational ("inter-viewpoint") inconsistencies.

We will discuss the activities involved in the "software engineering research lifecycle", including writing research grant proposals, hiring researchers, conducting the research, writing up and publishing the results and transferring them into industrial practice.

In this seminar we describe our work on a framework for monitoring goal oriented requirements specifications. The goals are specified formally using temporal logic. We will discuss how we automatically generate monitors, from the temporal logic specifications, to monitor the satisfaction or failure of goals. We will go on to discuss how we insert instrumentation into the monitored system to emit events to the monitors. This is done using AspectJ, an aspect oriented extension to Java, which we will briefly describe.

This talk introduces the probabilistic model checker PRISM being developed at the University of Birmingham. PRISM can be used to build a variety of probabilistic models (discrete- and continuous-time Markov chains and Markov decision processes). Models are specified using a simple state-based language, an extension of Reactive Modules. Two property specification languages are supported, PCTL (probabilistic computation tree logic) and CSL (continuous stochastic logic). PRISM is a symbolic model checker - its basic underlying data structures are BDDs and MTBDDs (multi-terminal BDDs), though for numerical computations three engines (sparse, MTBDD and hybrid) are supported.

In the talk will give a brief overview of issues in probabilistic model checking, outline the symbolic model construction together with selected algorithms, and finish with some examples. More information can be found on http://www.cs.bham.ac.uk/~dxp/prism/.

Recently, a number of reflective middleware platforms have emerged in an attempt to overcome such problems. This talk will report on the Open ORB reflective middleware architecture developed at Lancaster University. This architecture features the use of component technologies, component frameworks and both structural and behavioural reflection.

This architecture is currently being applied in a variety of aplication domains including mobile computing, programmable networks and the GRID. In the second part of the talk we will focus on the particular use of the technology in supporting mobile devices, in particular overcoming the high levels of heterogeneity that can be experienced in mobile and pervasive environments.

Dr. Emil Lupu is a lecturer in the Department of Computing at Imperial College. He has worked on policy-based network and security management for a number of years and is a co-founder of the Policy Workshop. His research interests include network and distributed systems management, distributed access control, security management and ubiquitous systems. Dr. Lupu was program co-chair of the Policy Workshop (2001) and of the IEEE Enterprise Distributed Object Computing Conference (2001). In addition, he has served on numerous program committees for conferences in network operations and management, access control and enterprise distributed systems.

The aim of the XMIDDLE project is the design and the implementation of a data sharing middleware for ad-hoc networks. The data structures that are stored by XMIDDLE are XML documents, which can be semantically associated with trees. The reconciliation process that we implement is executed transparently to users and can be easily adapted by developers according to their application requirements. In this seminar we will describe the data replication and synchronization techniques used to deal with possible data conflicts, using XML technologies, such as XML Schema.

There is a clear gap in research and practice in addressing the stability (or volatility) of the requirements of E-Type systems - at least in assessing this stability early in the requirements lifecycle. This presentation describes a new approach for early assessment of the stability of system requirements. The approach fuses two established techniques and the novel contribution is in the bringing together the existing work in Scenario Planning and Goal Directed Requirements Engineering. The presentation will describe its view of the problem of Requirements Stability, introduce Environmental Scenarios and illustrate how they can be linked with Goal Based Requirements approaches to address the problem. Using a NATS case study, the presentation will walk through the Goal Stability Assessment process using the Goal Stability Assessment Tools developed to support he process.

In this talk, we explain why real options theory appears to be well suited to predict architectural stability. We present ArchOptions, a novel model that exploits options theory to predict architectural stability. We show how we have derived the model from Black and Scholes options theory (Noble Prize Wining). we discuss: the analogy and assumptions made to reach the model, its formulation, possible interpretations, and usages. We discuss related and future work.

The key research challenge associated with the MDA is finding ways to transform models in a semantics preserving manner, for example from a PIM to a PSM. In my talk I will outline some ideas as to how these transformations can be described using the Object Constraint Language (OCL) at the UML meta-model level using the UML's lightweight extension mechanisms. I will also describe how some formal methods, particularly performance analysis, can be straightforwardly integrated into CASE tools supporting the MDA, by defining a semantic domain for the formalism, and describing mappings from designs to that domain.

We illustrate OGSA design features with reference to a fully productive e-Science application that we have modified to operate in a Grid environment using early manifestations of the Globus Toolkit 3.0. Computational chemists developed the original application combining functionality from several Fortran programs to create a process to predict stable crystal forms from an initial molecular structure. The process is extremely hungry for CPU time and even a simple analysis currently takes six to eight weeks to run. Apart from the limitation of running code on a single processor, scientists have to manually copy and reformat files so different parts of the process can run on platforms most suitable for a particular calculation. This lack of automation and resource coordination frequently results in errors forcing a total or partial rerun of the analysis. We show how an OGSA implementation facilitates the distribution of such applications across a large network of heterogeneous platforms radically improving performance of the system through parallel CPU capacity, coordinated resource management and automation of the business process.artial rerun of the analysis. We show how an OGSA implementation facilitates the distribution of such applications across a large network of heterogeneous platforms radically improving performance of the system through parallel CPU capacity, coordinated resource management and automation of the business process.

We investigate end-to-end quality of service (QoS) and highlight that QoS provision has multiple facets and requires complex agreements between network services, storage services and middleware services. We introduce SLAng, a language for defining service level agreements that accommodates these needs.

We illustrate how SLAng is used to specify QoS in a case study that uses a web services specification to support the processing of images across multiple domains in a quality of service aware manner. We evaluate SLAng based on the experience gained from this case study.

Peer-to-peer networks became popular quickly due to file sharing applications such as Napster or Gnutella. In these applications, people would let their computers cooperate, so that every user could download the other user's files, and vice versa.

Despite the unquestionable success, much still leaves to be desired by the actual service quality in file sharing networks. The download speed for a given file is usually far from the line maximum, and if it is, there is nothing a node can do to speed it up. The quality attributes of downloaded files, such as the level of compression or even just the completeness of the content, are also almost entirely unpredictable.

To stimulate cooperation in peer-to-peer networks, and both improve service performance and service reliability, we suggest a bartering based approach to provide incentives for nodes to offer services to other nodes. For this, we adapt a service oriented view of the peer-to-peer network. This allows the inclusion of other peer-oriented networks and services, such as Grid networks and message forwarding.

We will present the AdHocFS file system for mobile users, that allows collaborative data sharing among ad hoc groups that are dynamically formed according to the connectivity achieved by the ad hoc WLAN. It enhances, in particular, data availability within mobile ad hoc collaborative groups, and integrates a new adaptive data replication protocol for mobile terminals, combining both optimistic and conservative schemes in a way that minimizes communication and hence, energy consumption, while not restricting the nodes' autonomy.

Low-level biological behaviour is now being explored using 'biological modelling', a diverse family of techniques for representing and testing our understanding of biological systems and their emergent behaviour. Scaling up these models, through integration and combination, is a huge challenge, since each model may address a different aspect of a system, use different assumptions, or be based upon an entirely separate modelling paradigm.

The integration of models requires a detailed understanding of modelling itself. Therefore, in order to understand modelling we have constructed a meta-model, which we will eventually use as a means to integrate and scale-up models. In this talk, we will discuss our progress and preliminary findings.