October 2003 - September 2004
ESRC
Hydra is a grid-enabled spatial decision-support system. Its objective is to provide users with the ability to access and analyse data in order to make better decisions about the spatial allocation of resources, such as building new roads or redevelopment of brownfield sites in urban areas.
You can download a poster detailing the Hydra project here: Hydra Project Poster [PPT]
Spatial Decision Support Systems (SDSS) are a collection of technologies which are important to both academics and practitioners in a variety of real world environments. SDSS are closely related to Geographic Information Systems (GIS). GIS are software packages which allow the manipulation of spatial data ranging from satellite images to variations in household income. The capabilities of GIS include not only mapping at a variety of scales and styles, but also spatial queries (e.g. “show me the built-up areas”, or “where is the nearest hospital”) and some simple analytical functions (e.g. “how many burglaries were committed within two miles of Leeds University last month?”).
GIS can be used to address a wide variety of questions: for example, where should new places be made available in schools given recent trends in the birth rate in a city, or how will recent land use changes affect flood frequencies in a river valley? However, in order to provide adequate solutions to real problems, a number of additional capabilities are typically required. These capabilities include the provision of interfaces which are both easy-to-use and context specific; the ability to retain and compare user-defined scenarios; sharing of intelligence between the users of the GIS; and especially the provision of high value-added functionality, including spatial modelling and forecasting (Birkin et al, 1996). When GIS are bundled with other value-adding capabilities to support decision-making activities, the outcome is a spatial decision support system. SDSS are of interest to users in domains such as physical planning, local service delivery (schools, police, emergency services), health care, environmental monitoring and control, land use and transportation planning, and in business and commercial management, for example by banks and retail organisations (Geertman and Stillwell, 2003).
Both Spatial Decision Support Systems and Geographic Information Systems have been widely implemented within desktop computing environments. The emergence of the e-Science Grid provides an alternative infrastructure for the implementation of SDSS. ‘ Grid computing refers to the large-scale integration of computer systems ... to provide on-demand access to data-crunching capabilities and functions not available to one individual or group of machines’ (Foster, 2003). In essence, the Grid is seen as the computational equivalent of a National Power Grid. Just as one is able to plug a kettle into a power source and boil water using electricity, so in the next phase of computing it is argued that users will be able to access data and software with no need to worry about the location and setup of the hardware which supports these activities.
The fundamental aim of this project has been to develop a prototype, or demonstrator, of a grid-enabled spatial decision support system. The advantages of using a grid infrastructure in preference to a conventional desktop platform are considerable. In particular, it facilitates data integration at source, it promotes secure access and potential collaborative applications of decision support technologies, and it provides scalability to larger and more complex problems.
Hydra has been constructed around a scenario in which a specific health care service (of which a good example would be something like colorectal cancer screening) needs to be decentralised away from hospitals and into local clinics or GP surgeries. The decision support problem is to find an ideal solution which balances patient access with high quality provision and economic efficiency. The nature of the solution can be expected to vary for different kinds of solution – thus the ideal network for mother and baby clinics should look different to a screening programme for heart disease.
The software demonstrator incorporates the broad package of services which are necessary within a spatial decision support system. There are various data services, which allow users to access demographics, land registry statistics, health data and spatial data. The big advantage over conventional (desk-top) approaches is that as the data is updated on the provider sites, the new data automatically appears within Hydra, whereas this refreshment process is normally an expensive and cyclical process. Mapping services provide a variety of grid-enabled GIS functions. The modelling service allows ‘what if?’ scenarios regarding the placement of facilities to be configured and evaluated; and an optimisation service allows an ideal configuration to be identified. There is also a forecasting service which projects small area populations into the future. Finally, the reporting service provides relevant summary information from the various data sets and model runs.
These services have been developed using an Open Grid Service Architecture (OGSA) which essentially means that the latest e-Science technologies are being used to provide maximum system performance and security. In particular improved authentication means that it is now possible to offer this service to users without an account on the powerful White Rose Grid machines. In other words, users from all over the world can access our demonstrator application on machines located in Leeds, Sheffield or York.
In addition to the main prototype, we have also developed a secondary demonstrator – Hydra International – in which the same architecture is used to configure data access and interrogation from different census providers in four different countries. We have worked with our partners in the Worldwide Universities Network (WUN) in the UK, Europe and the US to articulate the composition and utility of a grid-enabled spatial decision support system for city planning around the world. Mark Birkin is a member of the ReDReSS Steering Committee and NCeSS Research Committee. We have presented our work at all of the e-Science All Hands Meetings, and at the first e-Social Science All Hands Meeting, and at a variety of international conferences and meetings.
In the future, we intend to work with NCeSS, University of Manchester, and with ReDReSS, University of Lancaster, regarding the dissemination of Hydra within the social science community. The architecture which we have developed for Hydra will be used as the platform for development of a wider range of research and policy applications for socio-demographic simulation and modelling. This work will take place from April 2005 under the umbrella of the MOSES (Modelling and Simulation for e-Social Science) which has been approved as a research node of the National Centre for e-Social Science. We are intending to build another ‘eCities’ demonstrator in collaboration with the Worldwide Universities Network (WUN). The eCities demonstrator will almost certainly build on the architecture of Hydra, and will most likely also re-use several of its individual software modules.
References