STATE OF THE PRACTICE TALKS 1

Session Chair : Michel Lemoine

Current Practices In Requirements Engineering: Where have we been and where are we now? More importantly, where must we go from here?

Donald C. Gause, Binghamton University

We are now up to the 12th IEEE International Requirements Engineering Conference. We can assume, rightly or wrongly, that eleven conferences have preceded this one and that many good ideas have been, and are continuing to be, introduced to enhance our understanding of Requirements Engineering (RE). We will take a look at these past twelve years to see just what has transpired. We may even take an occasional detour over the past several hundred years as we divert out attentions into more traditional areas of design and their "RE" approaches. Which approaches have worked, what has failed, and what can we learn from these for today's RE?

As we move into the present, we will look at our influence and successes in addressing important RE issues and opportunities. As the bottom line, how has RE improved the process of designing complex systems and how are we now able to develop more successful products - even in the face of increasing complexity and stakeholder diversity?

We quickly look to the future to outline a number of problem areas that we feel are rich in opportunity. These include such areas as: encouraging more active management involvement, improving our understanding of the value-added via various RE methodologies, closing the gap between RE research and RE practice, incorporating - rather than ignoring - uncertainty early in the design process, and identifying underlying fundamental principles of RE to avoid developing new methodologies every time we are posed with a new, unique problem domain (web-based computing, embedded systems, security-critical systems, synchronous/asynchronous systems).

Donald C. Gause is a Principal of Savile Row, LLC as well as a Research Professor in the Bioengineering Department of the Thomas J. Watson School of Engineering, State University of New York @ Binghamton. Don Gause has worked as an engineer and computer programmer and has managed engineering, programming and education groups with General Motors and IBM. He has been active as a consultant and professor for the past 35 years and served for many of these years as an adjunct member of IBM's Systems Research Institute (SRI). He has been a visiting scholar and has lectured at many universities and institutes around the world, has been an associate editor of the International Journal of Cybernetics and Systems, and has served as a national lecturer for a number of professional societies.

Mr. Gause's consulting and research interests include the development and analysis of requirements engineering and systems design processes, the design of user-oriented systems, and the management of innovation within large organizations. He has advised in the elicitation and documentation of business plans and requirements for Internet start-ups and Fortune 100 companies. He has also consulted on the development of strategic business systems, new products and processes for many leading firms. Mr. Gause is the author (with G.M. Weinberg) of Are Your Lights On?: How to Figure Out What the Problem REALLY Is, Dorset House, N.Y., 1990 and Exploring Requirements: Quality BEFORE Design, Dorset House, N.Y., 1989.

Requirements Traceability: Whither, Why and Wherefore

Dr. Jeremy Dick, Principal Analyst, Telelogic

For many organizations, traceability has become an end in itself. It is perhaps time to revisit traceability as an enabler for many requirements management processes. This paper briefly traces the history, surveys the state-of-the-art, and discusses the future of requirements traceability. It presents the latest thinking on the concept of "Rich Traceability", and discusses the vital role of traceability in change management. The conclusions are that there are some simple steps than can betaken by organizations to improve their processes through enhancing their use of requirements traceability.

Jeremy Dick obtained a DPhil from Imperial College in formal methods of software development, and pursued academic and industrial research in this field for a number of years. Since 1996, he has worked in the Telelogic UK professional services organization as a principal consultant in tool-supported requirements management. This role has afforded him a broad exposure to requirements management practices and issues across many industry sectors. He is co-author of the recent Springer book entitled "Requirements Engineering".

STATE OF THE PRACTICE TALKS 2

Session Chair : Martin Glinz

Requirements in the Rational Unified Process

Jim Heumann
Rational Software, IBM Software Group

This talk will provide an overview of the state-of-RUP with respect to requirements. Requirements play a central role within the Rational Unified Process (RUP). The talk will start with a brief history of the requirement discipline and how it got to where it is today. Next will come an explanation of how requirements are structured and used within the process. The presentation will also mention a variant called RUP for Systems Engineering and will highlight requirements within that framework. After that will come an explanation of some recent improvements in RUP as well a discussion of it's shortcomings. The talk will finish-up with some short usage stories highlighting how the RUP requirements discipline has been used by customers in various industries.

Jim Heumann is Requirements Evangelist for IBM Rational Software. Throughout his twenty-year career as a software designer, developer, maintainer, consultant and manager he as been able to experience, first-hand, most of the challenges we all face day-to-day as software professionals. He has worked on projects in artificial intelligence, fraud detection in cellular phone usage, user interface design, visual data analysis, currency trading and wind profilers, to name a few. Jim has delivered over fifty workshops, tutorials, training classes and conference presentations over the last ten years on a variety of subjects including Requirements Management with Use Cases, Object Oriented Analysis and Design, Visual Data Analysis, UML, and Using Use Cases to Generate Test Cases, among others. As his job title suggests, his primary role is to promote and educate on the importance of good requirements management. He holds an MS in Management Information Systems from the University of Arizona (USA).

Scenarios, Stories, and Use Cases in the Requirements Process

Ian Alexander, Independent Consultant

The requirements process, intended to solve 'the software crisis' and equivalent challenges in systems engineering, has traditionally focussed on documenting 'shall' statements as discrete individual elements. These proved hard to write and to interpret; hard to make complete and correct; but uncomfortably easy to make ambiguous and unstable given the extended life of many industrial contracts. This talk will explain how scenarios, along with related structures such as goals and viewpoints, are often the best remedies for these ills.

The basic idea of a scenario is a story, told from the point of view of one or several people who want to achieve a given result in the world. Stories are readily understood, with the advantage that anyone can tell them, they are not intimidating to non-specialists, and everyone knows how to spot errors and omissions in a story. There are many practical ways of telling engineering stories, from short chunks of text to perhaps the most familiar form, a numbered list of actions or steps, and on into the more elaborate shape of use case models, flow (activity) charts, and sequence diagrams which can be difficult for non-specialists to understand. Less verbal techniques can improve communication: acting-out scenes or stories in workshops; demonstrating sequences of interactions with mock-ups or prototypes; making slide-shows or video films to help people visualize using future products.

The talk will show how teams can use scenarios to discover and validate requirements; to guide the design and implementation of software and other products; to assist the selection of test cases; and even to select appropriate development life-cycle processes - which are themselves combinations of stories. The popularity of use cases, combined with the size of the software community, has meant that much of the thinking on scenarios has been concentrated on software development. This talk will suggest that the use case represents a part of the broad spectrum of scenario approaches, just as software forms a part of many kinds of system. It is wrong, however, to reason from the particular to the general: while every software/hardware product is a system, not every system conforms to the laws of software. This talk will illustrate how scenarios can help you to discover and document the requirements for your systems. It will show how you can use them to define what it will have to do, but also 'non-functional' requirements: both desired qualities (like reliability and security) and constraints (like cost and weight). Scenarios have been applied to businesses as diverse as automotive, air traffic control, telecommunications, railway, civil engineering, and software. This talk will show how scenarios can help you to build systems that meet the needs of your business.

Ian Alexander is an independent consultant specialising in Requirements Engineering. He is the author of the Scenario Plus toolkits. His book, 'Writing Better Requirements', is published by Addison-Wesley, 2002; his new book: 'Scenarios, Stories, Use Cases', is to be published by John Wiley, 2004. He has published numerous technical papers and is on the Program Committee of the Requirements Engineering 2004 conference. He is on the editorial boards of the Requirements Engineering Journal and the International Journal of Technology and Human Interaction. He helps to run the BCS Requirements Engineering Specialist Group and the IEE Professional Network for Systems Engineering. He is a Chartered Engineer.

STATE OF THE PRACTICE TALKS 3

Systems Engineering: A Vision for the Future

Regina Gonzales, Sandia National Laboratories

As with any endeavor it is vital to anticipate where the future of Systems and Systems Engineering is headed. One of the major objectives of the INCOSE is to create a Technical Vision of Systems and Systems Engineering in 2010 and 2020. Current Systems Engineering processes and practices must change in order to meet the demands of emerging systems. We are faced with enormous momentum of technology and demands on that technology. Since engineering in general, and Systems Engineering in particular, is the art and practice of enabling the realization of systems using technology, we must systems engineer our "Engineering". Besides the complexity and interconnectedness of systems today, we are faced with corporate infrastructure that is bogged down and a diverse and transitory workforce. In addition, many political, economic, social, technological, ecological, and legal constraints encumber the ability to realize systems effectively. This presentation will provide an overview of what INCOSE is doing to achieve the Technical Vision and the results of this project so far. Requirements Engineering plays a crucial role in Systems Engineering to first understand the problem and the application domains, and then formulating, managing and communicating a holistic specification of requirements for solutions.

Dr. Gonzales is currently a Systems Engineer and Project Manager in the Weapon Program Integration Department at Sandia National Laboratories. Her focus is on using modeling as a way to formalize requirements and requirements process. She is also an Adjunct Faculty member at New Mexico State University. Regina is the Communications Co-chair for the INCOSE Technical Board, the Co-chair for the Requirements Working Group, and the Chapter President of the New Mexico Enchantment Chapter. She has a PhD in Computer Engineering with a specialty in Requirements Engineering from New Mexico State University, an M.S. in Computer Science from University of Colorado, an M.S. in Electrical and Computer Engineering from University of Arizona, a B.S. in Electrical and Computer Engineering from New Mexico State University.

Safety requirements, safety claims and safety cases

Sofia Guerra, Adelard, UK

We all have heard tales about consequences of bad requirements: things go wrong, the budget and schedules overrun and business opportunities missed. The consequences of bad requirements are even more grave when the failure of the system may cause injury or death to human beings. The safety of such systems depends on the correct definition and realisation of the safety requirements. Safety requirements are those requirements of the functions intended to achieve or maintain a safe state of the system (IEC 61508). The justification for the adequacy of the safety requirements and their correct implementation should be based on a specific argument and a set of claims, supported by appropriate evidence.

In this talk I discuss the relationship between requirements, safety requirements and safety claims, and how these are combined into an argument about the safety of a system for a given application in a specific operating environment. Such an argument, supported by a body of evidence that provides a valid case that a system is safe is what it is called a safety case. Specific techniques for the elicitation and validation of safety requirements are considered, and their regulatory constraints set.

Sofia Guerra is a partner at Adelard, an independent, specialist consultancy founded in 1987. Adelard work is primarily in the area of computer based safety-critical systems with some important exceptions in our work on security, corporate memory and software reliability.

Since joining Adelard in 2000, Dr Guerra has been project manager of several research projects funded by the UK nuclear industry, both regarding the safety assessment and reliability estimation of software systems and in particular the assessment of smart sensor software. She has worked in a European project concerning the cost-effective modernisation of nuclear power installations, where she was responsible for developing its requirements engineering approach. Other work include the development of software for Internet-based calibration of DC voltage and resistance standards, the Health and Safety Executive research project on Incident Recording and Analysis research project, and the European Space Agency project to provide recommendations for complementing ESA's current approaches and standards concerning Software Product Assurance. She is working for the FlexRay consortium to review its safety work, including its safety concept and safety requirements and make recommendations on its Safety Management System. She is part of Independent Safety Advisor/Auditor team for several UK defence projects.

Dr Guerra has a degree in Mathematics and Computer Science (1989-1994) and a PhD in Mathematics (1999), both from Lisbon Institute of Technology, Portugal, and a BA in Music (1990). Prior to joining Adelard, she was a Research Fellow in the Department of Computer Science, University College London, where she was working on Requirements Engineering as part of the RENOIR network of excellence. She is a Member of the British Computer Society and a Chartered IT Professional.

STATE OF THE PRACTICE TALKS 4

Session Chair : Frank Houdek

Requirements Engineering Challenges for COTS Products

Xavier Franch, Universitat Politècnica de Catalunya (UPC), Catalonia

Commercial Off-The-Shelf (COTS) products can now be found in most business, government and defence organizations. Selecting, procuring and integrating COTS products are core activities of most systems development processes. Indeed COTS products are the predominant source of software for a wide range of today's applications -applications for which organisations and stakeholders still have requirements. However few existing requirements engineering methods, techniques and tools address COTS-based software development processes. Potential customers need requirements to select between candidate COTS products, write procurement contracts, guide COTS product integration and explore different product architecture configurations.

Some open issues are currently recognized as challenges for both the requirements engineering (RE) and COTS communities. Some of them were identified in the International Workshop on COTS and Product Software: Why Requirements are so Important (RECOTS), held jointly with the 11th RE Conference at Monterey (CA, USA) in September 2003, http://www.lsi.upc.es/events/recots/03/home.html. This open issues will be the core of the talk.
Deciding whether to COTS or not to COTS. Almost every system integrates COTS products at some extent, but not all the systems are COTS-based systems, i.e. systems mainly composed of several interconnected COTS products. COTS products are not a panacea in software engineering, they require some specific skills and infrastructure and remarkable system maintenance has its specific complexities (see below).
New role of requirements. Life-cycles vary in the presence of COTS products. New activities arise, such as COTS screening and selection, while others remain but may change. RE falls into this second sort. As pointed out in the rest of this summary, there are many differences concerning requirements when taking COTS technology into account and therefore RE processes should vary.
Source of requirements. Information about COTS products is an additional source of requirements in COTS-based systems. Among the topics related, it becomes necessary to identify information channels, and to know how trustable is the information available, specially when coming from COTS vendors.

Specificities of COTS requirements. It is a matter of discussion to determine if COTS requirements are different in nature than usual software requirements. Although the essence seems to be the same, differences exist. For instance, flexibility of requirements becomes a target, to fit to what's in the market; identification of non-negotiable requirements becomes then critical, as well as prioritization (also used for defining evaluation criteria). Generation of new requirements, discovered whilst examining the market, is more frequent than traditionally. Also, classification among core and peripheral requirements is helpful for COTS selection.
New types of requirements. Not only characteristics of requirements can be slightly different, but also some types of requirements are particular of the field. Remarkably, we can mention two new types. First, non-technical requirements, not only about the product but specially about the vendors or distributors. Second, requirements related to customisation and tailoring; some of them may be included in the contract.
Dealing with evolution. Software systems maintenance differs from the usual case, since a critical actor enters the scene, namely the COTS market. COTS products integrated in the system evolve, with new versions appearing; on the contrary, they may become obsolete and disappear. Competitor products may gain a better position and eventually could substitute the integrated ones. All these situations have an impact on system requirements. Last, it is not always easy to match market evolution and system evolution.

Xavier Franch is Associate Professor in the Software Department (LSI) at the Universitat Politècnica de Catalunya (UPC). He received his B.Sc. and Ph.D. in Software from the UPC. He is and has been a principal and co-investigator of several funded research projects. He is currently leading the GESSI group at the LSI (http://www.lsi.upc.es/~gessi/), compound of more than 10 full-time researchers. His current lines of research include requirements engineering, selection of COTS components, quality model construction and software process modeling. He has participated in several industrial collaborations of COTS selection in the fields of ERP systems, document management tools, health-care solutions and others. He has over 40 refereed publications in conferences, journals and books. In the areas related to the talk, he has publications in the IEEE International Conference of Requirements Engineering (RE), the IEEE Software magazine and the International Conference on COTS-based Software Systems (ICCBSS). He has won the best paper award in the 03 and 04 editions of the ICCBSS. He has
been program co-chair of the RECOTS'03 and MPEC'04 workshops related to COTS technologies.