LSS: London Software Systems

 

London Software Systems is a joint research institute established by Imperial College and UCL that focusses on techniques and tools for designing large-scale complex information technology systems.

Background & Rationale

Challenges

Operation

Director

Contact

Background and Rationale

Both Imperial College and UCL have research groups of outstanding international quality in the area of software systems engineering. The groups the Imperial College Distributed Software Engineering group (about 30 people, including academics, research assistants and research students) and the UCL Software Systems Engineering group (about 35 people, people including academics, research assistants and research students) have a long track record of collaboration and intellectual exchange.

 

Both groups have an excellent track record of obtaining EPSRC, EU and industrial funding founded upon a shared intellectual focus on engineering solutions which are lightweight, and are carefully targeted towards "real" industrial problems. Our approach is to develop prototype tools, which may include components from other groups and exploit emerging standards. These tools are made available to industrial and academic users for evaluation.

 

Our shared research approach is driven by case studies combined with rigorous analysis and validated in practice through collaboration and consultancy with industry. We are particularly aware of the need for software systems engineering techniques, methods and tools to scale and to be simple enough that they can be adopted in practice.

 

We have a common specialisation in the principled construction of large-scale potentially mobile, distributed systems and in the provision of infrastructure to support change, evolution and management in systems that may also have stringent quality-of-service requirements. We have a strong background in formal analysis and the development of tools and environments to support this. We have experience in developing tools for deployment and management of large-scale distributed systems across complex networks.

 

Our work links at one end to the strong Imperial College tradition of work on theory and formal methods as well as distributed programming environments and at the other end to the strong UCL tradition of work on networks and communications.

 

The establishment of London Software Systems reflects our common interests and track record. It is intended to raise further the international profile of the groups by providing a powerful shared 'brand' and identify us jointly as the leading software systems engineering group in Europe and one of the top 4 in the world. There is significant scope for shared support for research students and academic visitors and for extending our technology transfer profile. There is also scope for common computing infrastructure. It is also intended that LSS will, through complementary research expertise and strength in depth, leverage substantial additional research income.

Challenges

The context of software system development is changing. Systems are rarely developed from scratch; most system development involves extension of pre-existing systems and integration with 'legacy' infrastructure. These systems are embedded in complex, highly dynamic, decentralised organisations; they are required to support business and industrial processes that are continually reorganised to meet changing consumer demands and may have stringent quality of service and dependability requirements. The services that such a system provides must, for the life of the system, satisfy the requirements of a diverse and shifting group of stakeholders. There is a shift towards client and user centred approaches to development and an accompanying shift from a concern with whether a system will work towards how well it will work. Overall, fewer 'bespoke' software systems are being constructed. Instead, generic components are built to sell into markets. Components are selected and purchased 'off the shelf' with development effort being refocused on configuration and interoperability. The resulting systems are composed from autonomous, locally managed, heterogeneous components, which are required to cooperate to provide complex services. They are, in general, distributed and have significant quality-of-service constraints on their operation.

 

In this context the most pressing problem is the 'rigidity' of software systems set against the 'changeability' of the environment in which these systems must operate. This changeability ranges from context changes in mobile systems, at one end of the spectrum, to gross requirements changes, at the other end of the spectrum. The behaviour we require from the software systems in response to such changes is dependability and autonomous adaptation that preserves the desired system properties. For each position in the spectrum there are proposals or 'point' technological solutions but there has been no consistent or coherent attack on this problem. This constitutes the principal agenda item for the Institute.

 

Software systems engineering research is often best conducted in the context of the particular demands of challenging application domains. Such domains keep research 'grounded'. Areas in which the problems above are particularly evident and in which the groups have developed some expertise as well as application partnerships are transport, medicine, telecommunications and finance. Each domain has a set of common problems but also distinctive features. All can strongly build on interdisciplinary linkages within Imperial College and UCL.

Operation

It is envisaged that the Institute will operate as a virtual research centre anchored by a core laboratory resource. Staff and students will normally work in the 'home institutions' and will maintain their normal obligations to their 'home departments' but will spend periods working within the core laboratory facility. Such an arrangement is uniquely enabled by the physical proximity of UCL and Imperial College.

 

Project funding will underpin a programme of 3-6 month focused 'programmes' engaging staff and students. These programmes could draw on external participants drawn from industry or from other institutions as appropriate. The programmes would employ research hot-housing in which groups of individuals are brought together to work intensively on particular topics and that 'connected time' is carved out to support this. This mode of operation anticipates the 'Isaac Newton Institute' style of visitor lead programmes.

 

Other activities will be: a seminar programme; a joint programme for research students, including training and supervisory support; joint hosting of academic visitors; a shared technology transfer activity, including research brokerage and open briefing programme; venture incubation; provision of a shared software infrastructure.

Director

As a first step we are appointing a Director of London Software Systems at a Professorial Level. The goal of the Director will be to develop a research programme, build the Institute and to promote its activities.

Contact

For more details of LSS, please contact Anthony Finkelstein <a.finkelstein@cs.ucl.ac.uk> or Jeff Kramer <jk@doc.ic.ac.uk>