As a result of the 9/10 June meeting of the Engineering Research Commission, which discussed its Forward Look Programme, the EASE Programme will be subjected to further cuts over the next three years. A 30% cut from 1 April 1994 will reduce the effort on EASE from the previously agreed 13.5 staff years to 9.5 staff years. This cut will increase to 50% by 1 April 1996 when the effort level will be reduced to 7.5 staff years. ERC was not faced with any financial problems when it decided to make these cuts.
Unfortunately I am unable to give any details of the direct effects on the EASE work programme, as this has still to be determined by the ad hoc Panel which recommended the cuts. The Panel is of the view that the EASE Programme should be directed to meet the needs of the ERC Committees. The Panel will decide what these needs are when it determines the activities under the reduced programme. I hope to be able to spell out how the cuts will affect you, the user community, in the next issue.
The IKBS Bulletin Board is a moderated, electronic mailing list provided under EASE and run by RAL's Knowledge Engineering Group. Subscription to, and use of, the service is free.
The purpose of the bulletin is to provide links between several academic and industrial interest groups. These include researchers into intelligent systems and AI, systems engineers developing leading edge applications, the cognitive science community and others.
The mailing list contains approximately 200 individual addresses, relating to approx 100 institutions. Many of these are, or act as, local nodes so that the actual audience is likely to be considerably larger.
The readership is international and world-wide, which is evident from requests we get, either for inclusion in the mailing list, or for publicising events, etc. Through the use of a moderator, information exchange within the communities is achieved with very low noise (avoiding the chatter box problem).
Each issue of the bulletin carries information under the following headings: Announcements (to advertise courses, conferences, workshops or similar), Calls for Papers (for conferences or journals, eg special issues), Correspondence (for brief information on research results or requests for collaboration), and last but not least, a heading JOBS to advertise openings in academia or industry.
Issues vary in length but are usually between 300 and 800 lines (ASCII). Individual items when submitted should be less than 100 lines of text.
f you/your organisation would like to receive the IKBS Bulletin and/or want to submit information to be broadcast then send an E-mail to:ikbsbb@uk.ac.rl.inf.
The following grant report was funded by the Process Engineering Committee.
Bio-SCAN (Bioprocess Supervision Control and Analysis) is a real-time KBS development using the G2 Real-Time Expert System toolkit (Gensym Corp) which has been designed to provide a general purpose framework for the monitoring and supervisory control of a range of fermentation processes. Internally, the system incorporates knowledge of operating strategies (eg feeding regimes, process set-points) and expected behaviours to enable rapid detection of any process faults. When faults occur, the system attempts to diagnose the cause of the problem and will offer recovery advice to process operators. In addition to knowledge encoded within Bio-SCAN, the utility of the system has been extended by interfacing to conventional algorithmic methods of analysing and controlling fermentations. A toolkit of on-line modules has been developed to incorporate a number of features into the overall control strategy such as simulation and prediction, statistical evaluation and estimation algorithms.
The proposed arrangement for such a system is shown schematically in Figure 1, comprising of a fermentation supervisory knowledge base (BioSCAN) and a separate Plant Scheduler knowledge base. Both KBS are supported by an on-line relational database (RDB) to provide a high level of integrity and 'intelligence' in a real-time environment.
The basic philosophy behind the development of Bio-SCAN has been to make the solutions as generic as possible (applicable across a broad spectrum of fermentations) and to provide the system with as much (high quality) information as possible. This led to the development of a database system for on-line interrogation by Bio-SCAN.
The important problems to be tackled when developing fermentation databases are not merely how to collect large volumes of different types of data, but also how to validate it before storage and then extract meaningful information. The system chosen to develop our database was Ingres by Relational Technology Inc. which operates a powerful relational database management system. Ingres proved simple to design and update via its forms-based-interfaces and supports a powerful query language (SQL) based on virtually any combination of criteria. SQL can also be embedded in external host code such as C which proved important in establishing the link with the scheduling and supervisory knowledge bases.
Bio-SCAN is linked to the pilot scale fermentors in the University of Newcastle's fermentation unit via a high speed fibre-optic ethernet connection to an IBM AT computer. The AT simulates an industrial plant data logging computer and receives process information through an RS232/422 serial connection from number of industrial TCS (Turnbull Control Systems) signal processors and controllers which are in turn directly linked to the process. The passage of data in or out of the G2 environment is accomplished using GSI, the G2 Standard Interface. GSI reads a process data file written by the AT at approximately 1 minute sampling intervals.
To enable a flow of information between supervisory and scheduling knowledge bases, running on separate machines, we have used Gensyms G2 to G2 interface. This interface runs on top of ICP (Intelligent Communications Protocol) which is also used for the operation of GSI, mentioned earlier.
For communication between KB and database an interface module is required, which is essentially composed of two elements, a Gateway and a Server. The Server is designed to implement all the database functions (requests for data retrieval or data storage) that are needed by each KB, while the Gateway implements a transparent interface between the KB and the Server using GSI. When a request is made from a KB, it passes via the Gateway to the Server, accompanied by certain timing information. This allows the Gateway to monitor the progress of the Server in processing the request. If the Server fails to complete the request within a specified time, the Gateway will send back an error message to the KB. If the request is successful, information will either be retrieved from the database and sent back through the gateway or, in the case of a storage request, information will be written to the database and a completion code will be returned to the Gateway. The database Server implements a stateless protocol for query and storage commands; each request made to the Server contains all the information the Server needs to successfully complete the request.
The separation of the interface into a Gateway and a Server allows for distribution of these modules over separate, dedicated processors so that database operations do not impair the constraints associated with real-time KBS operation. This is particularly important for the supervisory KBS.
Communications to the database Server are not limited to the two KB's alone. Other on-line modules associated with Bio-SCAN and the scheduling forecaster must also have access to process information. In addition, the database itself must transfer historical process information to some mass storage medium (eg magnetic tape) and retrieve this data in response to user queries. For this purpose a library of functions have been written which will allow any process to establish a link with the Server in a manner similar to that described for the Gateway operation. Off-line data from laboratory analysis must also be placed in the database. This could be input manually by an operator using an Ingres forms-based interface which prompts for the information needed, or semi automatically from a data file. The complete system which is being developed for real-time operation is shown schematically in figure 2. The two knowledge bases and the database run on SUN 3/60 workstations each with 12 MBytes of memory and individual 140 MByte disks. Each is connected in a LAN that includes a I km link to the fermentation unit.
Although we are currently at an intermediate stage in the project, the potential benefits of the proposed system appear quite significant. In large scale, established fermentation processes such as penicillin production, or in smaller scale value-added fermentations such as vaccine production, even a small increase in fermenter or plant productivity will realise major savings. Bioprocess and Pharmaceutical companies, such as Eli Lilley in North America, are already taking advantage of RTKBS techniques. Other companies involved in processing, such as sugar producers in Italy, Unilever world wide and industrial members of the UK Cogsys Club are already engaged in applying the concepts described.
The authors would like to thank Smith-Kline Beecham and Gensym Corporation for valuable assistance. We also acknowledge financial support provided by the SERC.
First-rate research in UK universities needs to make an impact on those who will benefit most. The Science and Engineering Research Council's Construction Committee has taken the initiative in bridging the information gap by appointing Joan Van Emden, formerly Editor of the SERC newsletter Research in Building, to the new post of Building Research Information Coordinator.
Her aim is to take the very best research sponsored by the Committee and, through the leading construction journals, to make the results available to the industry. Recent outstanding research has ranged from cladding systems and the problems of airborne and structural vibration, to automation and robotics and a new approach to the training of architects.
My aim is to highlight the excellent research which our universities consistently produce, says Joan van Emden, and to let the industry know about new systems and products which will revolutionise the building site - and the manager's office - into the twenty-first century. The new post, funded initially for one year, is launched this month.
FEUSERS is a mailbase-list serving the researchers, teachers and groups active in using and developing FINITE ELEMENTS and FINITE DIFFERENCE software. Code developers and vendors of finite element and finite difference packages are particularly encouraged to be involved in FEUSERS. Users and developers of graphical pre- and post-processing systems (with solid modelling) and CAD/CAM interfaces for finite elements and finite difference applications software are also encouraged to participate. Industries and industrial participants with a solutions approach would be highly appreciated at this forum.
The FEUSERS forum aims to encourage discussion and free exchange of ideas amongst the members of the list. As a list member, you are free to raise a topic and express your views on it, invite other views and even summarise them. You can also respond to a topic currently under discussion.
All postings to the group are distributed to all the list-members, and the list will not initially be moderated.
Discussion topics on finite element and finite difference applications areas such as structural and mechanical engineering, analysis and design, aerospace engineering, electronics and electrical engineering, computational fluid dynamics, heat and thermal analysis, impact analysis, chemical engineering, nuclear and solid state physics, etc. are highly desirable. Specific suggestions and questions on various packages (e.g. ideas, patran, abaqus, nastran, lusas, emas, fidap, phoenics, etc) could be beneficial for many finite element and finite difference users and developers.
The forum encourages notices and discussion on finite element and finite difference references, books, journals, publications, information, projects, collaborations, proposals, funding, hardware, software, organisations, visits, shows, standards, education, research, development, jobs, conferences, surveys, etc.
To JOIN the list and become a list-member of feusers, E-mail the following line:
JOIN feusers <YOUR_Firstname> <Your_Surname>
to mailbase@uk.ac.mailbase. To send in YOUR CONTRIBUTION to the list, E-mail the item to: feuserS@uk.ac.mailbase.
The latest agreements from CHEST include:
Forthcoming agreements include:
How many readers have been given a tapeful of images for some important work, only to find that no information has been supplied about the tape format? In such cases, the effort involved in ensuring that the tape has been read correctly and converted into one's own file format is usually at least as great as the effort involved in processing it. Equally irritating, how many times have you received a program from someone, only to find it uses an image processing library which you do not have? You might be surprised --- and hopefully pleased to learn that many of these problems could disappear in the near future due to work on an international standard for image processing and interchange.
First, the bureaucratic aspects. The effort is ISO project 1.24.10, and is taking place under ISO/IEC JTCI SC24 WG7. The terms of reference of SC24 are Computer Graphics and Image Processing, and they have produced standards such as GKS, CGM, and PHIGS in the past. The name of the project is Image Processing and Inter-change (IPI), and subsumes earlier ANSI activity known as PIK (more of which below). The developing standard will be published as IS-12087. IPI comprises three parts:
Note that these documents describe functionality only, independent of any programming language or interchange mechanism. To map on to a programming language, a language binding must be used; similarly, an encoding must be used to map the interchange mechanism on to something physical.
One language binding is currently under way, in ISO C, being prepared by Jerry Paquette, an independent consultant in the USA. A Fortran 90 language binding was planned but will probably not reach fruition due to the lack of a document editor (any volunteers?). There is also significant interest in preparing a C++ language binding, if the C++ standard ever happens (if it doesn't, we'll produce a set of guidelines). The language binding will be published as a separate standard.
The primary aim of IPI is obviously to provide a widely-accepted standard for both processing and interchange. To ensure that the processing and interchange parts work together, the Part 1 document presents common introductory and architectural material, such as the data types (both elementary and structured) that are used in both of the other parts, the processing model, and various ground rules (restrictions). It also outlines the method by which conformance will be achieved.
The Part 2 document describes the processing part of the standard, known as IPI-PIKS. This is derived primarily from the earlier ANSI work called "PIK", but the processing model and range of operations available have been greatly elaborated. PIKS images are (at the highest level of conformance) represented as S-D objects, the dimensions nominally being three spatial, one time, and one spectral. There are some rules regarding the data types that may be used and the ordering of the five dimensions, but it's not worth going into these here. PIKS provides (if my memory serves me right) about 2000 operators, many of which have several options: for example, the "dyadic" operator can add, multiply (etc) images. The coverage of the operators is pretty much what you would expect. One important thing to note is that only pixel-processing operations are considered: in many application areas (eg, vision), pixel representations are quickly left behind; IPI will not support the non-pixel-processing parts of such applications. The Part 2 document also describes facilities for loosely aggregating S-D images into higher-level constructs, such as image lists.
A further important aspect of IPI-PIKS is the ability (in the highest levels of conformance, at least) to allow application programs to execute operators asynchronously. Furthermore, a series of operator calls may be bundled into a thing called an "operator chain" (a macro, in effect) which may itself be executed asynchronously.
The interchange part of IPI, called IPI-IIF, aims to provide facilities that will subsume those offered by de facto standards such as TIFF and TIFF/IT (which most people already know of), ACR-NEMA (for medical imaging), and FITS (for astronomical data). IPI-IIF is based around a formal syntax in a sort of son-of-BNF known as ASN.1 (an OSI layer 6 thing, though IPI-IIF actually lies outside the OSI layer structure). It will provide extensive facilities for including structure into the image datastream and can transfer various types of annotation (including graphic and audio) but will not be as all-encompassing as, say, ODA. Image-related data types such as regions-of-interest, histograms, etc, will also be capable of transfer via IPI-IIF. Standardized compression algorithms (eg, JBIG, JPEG, MPEG, and the fax standards) will be available for the compression of image data. Quite a lot of effort has also been expended in the area of colour: if colour images are to be transferred via IPI-IIF, they may be specified in any widely-accepted colour representation (and some that aren't), with conversions between representations possible via IPI-PIKS.
Just as IPI-PIKS may have separate language bindings, IPI-IIF may be encoded in different ways. One encoding is currently being developed, namely an ASN.1 encoding. This is generating a certain amount of concern at the moment, since "vanilla" ASN.1 (the basic encoding rules, for those in the know) are unsuitable for sending large numbers of pixels. The ASN.1 people are working on a set of encoding rules for packed (eg, pixel) data which may be mature enough to be adopted for IPI; if not, a separate IPI-TIP packed encoding will be produced.
My personal opinion about the IIF is that it is unlikely to gain acceptance as a common file format within an organization or site, but that it may well become the norm for interchanging images between organizations or sites. In particular, there is interest in adopting IPI-IIF as the basis of a future interchange format within the medical imaging community.
One particular area that has been tightened up over the last year is that of conformance. There are now well-defined lists of data types, operators etc that must be supported by implementations. A scheme for specifying the numerical properties of data representations and arithmetic has also been put into place it would be very unfortunate if, say, someone implemented a Fourier transform that output only bi-level values!
Having described all this, the main question in your mind will probably be When will it all happen? Well, the IPI documents are currently half-way through their Draft International Standard ballot (which closed in June). Any comments generated by national standards bodies will be discussed at a meeting in Colorado, USA in early July. At that point, it is feasible that the documents could be recommended for progression to International Standard, though I would guess that a second (three-month) DIS round might be necessary. All being well, the International Standard would follow within a year. (Note that the IPI work only started in 1991, so we getting to a standard in record time - think how long GKS took to standardize!)
Copies of the DIS text should be available via your national standards body. Due to copyright restrictions and the thickness of the documents (over 2.5in), I cannot send copies to people, so don't ask! However, I will do my best to answer any genuine queries that people may have.
The Computational Fluid Dynamics Community Club is part of the Engineering Applications IT Support Programme. It was founded in late 1989 and now has a registered membership of 500 academics and industrialists representing more than 65 higher educational institutes. Three of its objectives are to increase the awareness of advanced computing and software engineering techniques in the community, to encourage an exchange of views and information, and to promote the exchange of CFD software and data throughout the community. One of the requirements of this Community Club was easier access to commercial CFD software on supercomputers. To satisfy this need we have negotiated special licences with a number of commercial software vendors to allow academic use of their products on the Cray Y-MP/8 at the Atlas Centre.
The packages involved are:
The arrangement for the first four packages is that existing academic holders of workstation licences can use the same software on the Cray without obtaining a separate, usually more expensive, Cray licence. The fifth package, FLOW -3D, is freely available to all academic users. Although there are no software costs involved, a Research Council grant is required to cover the CPU time used on the Cray. A small amount of CPU time (5 Y-MP CPU hours over three months) can be obtained through a Pump Priming award in order to evaluate the software. Pump Priming awards can be obtained quickly, but only by those eligible to apply for Grants (tenured year academics at UK HEIs) or Research Council staff.
For more details of the CFD software, contact me.
The Parallel Processing Special Interest Group (PPSG), founded in 1986, exists to foster the development of parallel architectures, languages and applications and to disseminate information on parallel processing. Membership is completely open; you do not have to be a member of the BCS to join.
All members receive the twice yearly issues of our newsletter, Parallel Update, and of our Bulletin, Parallel Lines. These advertise and record the activities of the group and advertise conferences and meetings of interest to members. Reviews of specialist literature are provided and special features summarise trends in high performance computing. In addition, members get substantial discounts to PPSG events (such as those below) and to those organised by other BCS Specialist Groups and commercial conference organisers.
The PPSG has arranged lectures on: The MasPar MPI, TAOS, The Intel TouchStone Project - The Intel Paragon, The CM-5, The Alliant Campus. PPSG Conferences and Workshops have included: Affordable Parallel Processors, Languages for Parallel Processing, Parallel Architectures and Computer Vision, Compiling Techniques and Compiler Construction for Parallel Computers, Abstract Machine Models for Parallel Computers (with the University of Leeds).
There has been an encouraging response from engineering researchers from both academia and industry to the launch of the new Parallel Processing in Engineering Community Club which was announced in the May Issue of ECN. Membership has grown rapidly to over 70 in its first two weeks.
The aims of the Community Club are:
If you are interested, please contact us to register as a member. A brief questionnaire will be sent to Club members to gather further information about the current parallel processing activities within the engineering research community. Send a message including your name, address, telephone and E-mail details to ppecc@uk.ac.rl.inf.
Many data processing systems fail to meet the needs of the end-users. Poorly specified user requirements and changing requirements are two of the main reasons. Data Modelling and Database Design aims to address exactly this problem, and the seminar should equip the attendee with a knowledge of how to analyse the business, scientific or engineering requirement and how to fit the information system design compatibly with this need. Key aspects include designing for flexibility and generality, and the key factors in gaining security and performance.
Programme:
COMETT Seminars on Database Management: A series of 10 seminars on Database Administration - Problems, Concepts and Issues -is currently being planned to run from September to December 1993. The series is funded under the CEC COMETT Programme and its goal is to produce highly trained and qualified Database Managers in order to improve the competitiveness of European Industry. It is therefore mainly targeted at Industry however there will be a number of places available to academics.
The seminars have been developed by the University of Hagen (FemUniversitat), the German equivalent of the UK's Open University, and have already run successfully in Germany. Rutherford Appleton Laboratory is a partner in the project and is the UK sessions Coordinator. Some of the topics to be covered are:
The series of seminars is developed in such a way that participants may attend all events or select those particularly relevant to their area.
