ESPOO 2004 VTT PUBLICATIONS 526

VTT PUBLICATIONS 526

VTT PUBLICATIONS
509 Sallinen, Mikko. Modelling and estimation of spatial relationships in sensor-
based robot workcells. 2003. 218 p.
510 Kauppi, Ilkka. Intermediate Language for Mobile Robots. A link between the
high-level planner and low-level services in robots. 2003. 143 p.
Juho Jäälinoja
511 Mäntyjärvi, Jani. Sensor-based context recognition for mobile applications.
2003. 118 p. + app. 60 p.
512 Kauppi, Tarja. Performance analysis at the software architectural level. 2003.
78 p.
Requirements implementation in embedded
software development

Requirements implementation in embedded software development

513 Uosukainen, Seppo. Turbulences as sound sources. 2003. 42 p.
514 Koskela, Juha. Software configuration management in agile methods. 2003.
54 p.
516 Määttä, Timo. Virtual environments in machinery safety analysis. 2003. 170
p. + app. 16 p.
515 Palviainen, Marko & Laakko, Timo. mPlaton - Browsing and development
platform of mobile applications. 2003. 98 p.
517 Forsén, Holger & Tarvainen, Veikko. Sahatavaran jatkojalostuksen asettamat
vaatimukset kuivauslaadulle ja eri tuotteille sopivat kuivausmenetelmät.
2003. 69 s. + liitt. 9 s.
518 Lappalainen, Jari T. J. Paperin- ja kartonginvalmistusprosessien mallinnus ja
dynaaminen reaaliaikainen simulointi. 2004. 144 s.
519 Pakkala, Daniel. Lightweight distributed service platform for adaptive mobile
services. 2004. 145 p. + app. 13 p.
520 Palonen, Hetti. Role of lignin in the enzymatic hydrolysis of lignocellulose.
2004. 80 p. + app. 62 p.
521 Mangs, Johan. On the fire dynamics of vehicles and electrical equipment.
2004. 62 p. + app. 101 p.
522 Jokinen, Tommi. Novel ways of using Nd:YAG laser for welding thick section
austenitic stainless steel. 2004. 120 p. + app. 12 p.
523 Soininen, Juha-Pekka. Architecture design methods for application domain-
specific integrated computer systems. 2004. 118 p. + app. 51 p.
525 Mäntyniemi, Annukka, Pikkarainen, Minna & Taulavuori, Anne. A Frame-
work for Off-The-Shelf Software Component Development and Maintenance
Processes. 2004. 127 p.
526 Jäälinoja, Juho. Requirements implementation in embedded software devel-
opment. 2004. 82 p. + app. 7 p.

Tätä julkaisua myy Denna publikation säljs av This publication is available from
Juho Jäälinoja
VTT TIETOPALVELU VTT INFORMATIONSTJÄNST VTT INFORMATION SERVICE
PL 2000 PB 2000 P.O.Box 2000
02044 VTT 02044 VTT FIN–02044 VTT, Finland
Puh. (09) 456 4404 Tel. (09) 456 4404 Phone internat. +358 9 456 4404
Faksi (09) 456 4374 Fax (09) 456 4374 Fax +358 9 456 4374

ISBN 951–38–6370–0 (URL: http://www.vtt.fi/inf/pdf/)

ISSN 1455–0849 (URL: http://www.vtt.fi/inf/pdf/)
 VTT PUBLICATIONS 526

Requirements implementation in
embedded software development

Juho Jäälinoja
VTT Electronics
ISBN 951–38–6370–0 (URL: http://www.vtt.fi/inf/pdf/)
ISSN 1455–0849 (URL: http://www.vtt.fi/inf/pdf/)
Copyright © VTT Technical Research Centre of Finland 2004

JULKAISIJA – UTGIVARE – PUBLISHER

VTT, Vuorimiehentie 5, PL 2000, 02044 VTT
puh. vaihde (09) 4561, faksi (09) 456 4374
VTT, Bergsmansvägen 5, PB 2000, 02044 VTT
tel. växel (09) 4561, fax (09) 456 4374
VTT Technical Research Centre of Finland, Vuorimiehentie 5, P.O.Box 2000, FIN–02044 VTT, Finland
phone internat. + 358 9 4561, fax + 358 9 456 4374

VTT Elektroniikka, Kaitoväylä 1, PL 1100, 90571 OULU

puh. vaihde (08) 551 2111, faksi (08) 551 2320
VTT Elektronik, Kaitoväylä 1, PB 1100, 90571 ULEÅBORG
tel. växel (08) 551 2111, fax (08) 551 2320
VTT Electronics, Kaitoväylä 1, P.O.Box 1100, FIN–90571 OULU, Finland
phone internat. + 358 8 551 2111, fax + 358 8 551 2320

Technical editing Marja Kettunen

Jäälinoja, Juho. Requirements implementation in embedded software development. Espoo 2004.
VTT Publications 526. 82 p. + app. 7 p.
Keywords software process improvement, software requirements analysis, embedded
systems

Abstract
Development of correct requirements at the beginning of a software project is
considered an important precondition for successful software development.
Moreover, implementing these requirements correctly during the software
development is arguably just as important. Rigorous implementation of
requirements in embedded software development is especially critical, since
requirements affect both software and hardware. The goal of this research is to
identify elements for effective requirements implementation in embedded
software development.

A conceptual-theoretical research approach is applied to analyse previous

research on requirements implementation and to construct a new theory which
integrates requirements implementation related elements into a holistic
framework. These elements include requirements implementation processes,
methods, and roles. The developed framework describes relations among these
elements and furthermore their relation to software development activities. The
framework can be used as a basis for improving software development areas that
are related to requirements implementation.

To validate the feasibility of the developed framework, two case studies were
carried out within embedded software development organisations. The validation
was conducted by making a current state analysis and by suggesting
improvements based on the developed requirements implementation framework.
The results from the case studies indicated that the framework was a useful
foundation for improving the organisations' requirements implementation
practices.

3
 Preface
This research was conducted at VTT Electronics in the MOOSE (software
engineering MethodOlOgieS for Embedded systems) project. The project's goal
was to improve the integration of available tools, techniques, methods, and
processes involved in embedded software development. The project started at
the beginning of 2002 and I joined the project a year later. In the project I had an
opportunity to study and write my graduate thesis on requirements and
embedded software development. This publication is based on that thesis. I must
sincerely thank VTT Electronics and other partners in the project for allowing
me to write the thesis as part of the project.

I wish to thank all the people at VTT Electronics for their support. Especially, I
would like to thank Ms. Päivi Parviainen for competent guidance throughout the
study. Furthermore, I want to thank Prof. Markku Oivo from the University of
Oulu for his support. I am also grateful for the co-operative people in the two
anonymous organisations I had the opportunity to study.

Oulu, March 2004 Juho Jäälinoja

4
 Contents

Abstract................................................................................................................. 3

Preface .................................................................................................................. 4

Abbreviations........................................................................................................ 7

1. Introduction..................................................................................................... 9
1.1 Background............................................................................................ 9
1.2 The research problem and methods ..................................................... 10
1.3 Scope ................................................................................................... 12
1.4 Structure .............................................................................................. 12

2. Embedded systems and requirements ........................................................... 14

2.1 Embedded systems .............................................................................. 14
2.2 Embedded systems development......................................................... 15
2.3 Embedded software development........................................................ 16
2.4 Requirements....................................................................................... 18
2.5 Requirements engineering ................................................................... 21

3. Requirements in software development activities ........................................ 24

3.1 Software requirements analysis ........................................................... 24
3.2 Requirements and software architecture design .................................. 26
3.3 Requirements and detailed software design ........................................ 29
3.4 Requirements and software coding...................................................... 30
3.5 Requirements and software testing...................................................... 31

4. Requirements implementation-supporting elements..................................... 34

4.1 Requirements change management ..................................................... 34
4.1.1 Impact analysis........................................................................ 36
4.1.2 Change management tools ...................................................... 37
4.2 Requirements tracing........................................................................... 38
4.3 Consistency management .................................................................... 41

5. Requirements implementation framework.................................................... 45

5.1 Requirements implementation framework for embedded software..... 45

5
 5.1.1 Requirements implementation in software requirements
analysis.................................................................................... 46
5.1.2 Requirements implementation in software designing ............. 50
5.1.3 Requirements implementation in software coding.................. 52
5.1.4 Requirements implementation in software testing .................. 54
5.1.5 Requirements implementation throughout the development... 56
5.2 Adaptation of the framework............................................................... 59

6. Validation of the requirements implementation framework ......................... 62

6.1 Research approach for validation ........................................................ 62
6.2 Case one - large organisation............................................................... 63
6.2.1 Current state analysis .............................................................. 64
6.2.2 Improvement proposal ............................................................ 65
6.3 Case two - small organisation.............................................................. 67
6.3.1 Current state analysis .............................................................. 67
6.3.2 Improvement proposal ............................................................ 68
6.4 Applicability of the requirements implementation framework............ 69

7. Conclusions................................................................................................... 71
7.1 Answers to the research questions....................................................... 71
7.2 Significance of the results ................................................................... 72
7.3 Further research possibilities ............................................................... 74

References........................................................................................................... 75

Appendices

Appendix 1: Questionnaire on current state requirements implementation

practices
Appendix 2: Requirements implementation practices

6
 Abbreviations
AQA Architecture Quality Analysis
ASIC Application-Specific Integrated Circuit
ATAM Architecture Tradeoff Analysis Method
CBSP Component-Bus-System-Property
CCB Change Control Board
CMM Capability Maturity Model
CMMI Capability Maturity Model Integration
DOORS Dynamic Object-Oriented Requirements System
DSP Digital Signal Processing
GRL Goal-oriented Requirement Language
MOOSE software engineering MethodOlOgieS for Embedded systems
PDL Program Design Language
QFD Quality Function Deployment
RTM Requirements Traceability Management
SBAR Scenario-Based Architecture Reengineering
SCM Software Configuration Management
SCR Software Cost Reduction
SDL Specification and Description Language
SRS Software Requirements Specification
TCM Test Case Management
UCM Use Case Maps
UML Unified Modeling Language
VTT Technical Research Centre of Finland
(Valtion teknillinen tutkimuskeskus)

7
 1. Introduction
Modern embedded systems are more complex than before, and much of this
complexity is realised in these systems' software. Organisations that neglect to
manage software development efficiently are doomed to fail in these complex
development projects. Developing correct requirements and implementing them
correctly are considered key elements in successful software development. In
this research, the latter condition is studied and related means for improving
requirements implementation are provided. The term requirements
implementation refers in this research not only to realising requirements in code,
but also to their elaboration into software requirements, designs, and test cases.
This chapter presents the background for the study, defines the research problem
and scope, and outlines the structure of the study.

1.1 Background

This study was carried out at VTT Electronics within the MOOSE (software
engineering MethodOlOgieS for Embedded systems) project, which aimed to
improve the integration of available tools, techniques, methods, and processes
involved in embedded software development. System and software requirements
engineering are among the research areas in the project. The industrial partners
of the project expressed several problems related to requirements. The following
list contains some examples (MOOSE 2002b):

• requirements are hard to specify and difficult to know up-front

• software expertise is absent from system requirements allocation
• change of requirements creates inconsistency amongst work products
• tracing of requirements throughout the development is difficult

One of the main problems of requirements implementation is the inevitable

change of original requirements. Requirements may be removed, modified, or
added to during the development. Software development is very rarely a linear
front-end process, where requirements can be agreed upon at the beginning and
the rest of the development will be based on them. Instead, software typically

9
iterates towards greater maturity. During the iterations original requirements
very often change, which can cause problems for design, implementation, and
testing activities. Problems typically arise when a change of one requirement
affects many different components in the system.

Correct implementation of requirements in embedded software development is a

critical success factor for the whole system development since it affects both
software and hardware. For instance, an inconsistency between requirements and
software designs leads to incorrect implementation of the software and
furthermore might trigger unexpected behaviour in the system's hardware. The
later a defect is found in the system development, the more it will cost to repair.

This research explores solutions to these problems from the point of view of
requirements implementation. Although there has been made a significant
number of research on development, implementation, management, and
consistency management of requirements, these studies unfortunately present
only insular solutions. A holistic view of these elements is needed in order to
locate essential improvement areas in software development. Furthermore,
current research lacks an embedded software viewpoint regarding these issues. It
is obvious that concurrent development of hardware and software creates
specific problems for requirements implementation. In this study, the previously
stated deficiencies in current research are addressed.

1.2 The research problem and methods

The purpose of this research is to study effective implementation of

requirements in embedded software development. The research problem is stated
as the following research question and sub-questions:
1. How can requirements be effectively implemented in embedded software
development?
1.1. What is the relationship between requirements and
development activities?
1.2. How should requirements be implemented during
development?

10
 1.3. What kind of framework would help to analyse and improve
requirements implementation practices?

The research problem is solved with conceptual-theoretical and theory-testing

research approaches. Conceptual-theoretical research approach usually includes
an analysis phase, which is sometimes followed by a synthesis phase. The
objective of the analysis phase is to study how previous research has structured
the subject of research. (Järvinen & Järvinen 2000.) The first research sub-
question is answered here in the analysis phase by studying the relationship
between requirements and development activities based on current literature.
Literature is also used as a source for answering the second sub-question.
Elements for effective requirements implementation are assembled from
software engineering standards, books and research papers.

The synthesis phase of conceptual-theoretical research approach allows us to

construct a new model or a theory, which describes the subject of the research
more accurately (Järvinen & Järvinen 2000). In the synthesis phase, the third
research sub-question is answered by integrating requirements implementation
related elements found in the analysis phase into a holistic framework. The
developed framework describes relations among requirements implementation
elements and furthermore their relation to software development activities. The
utilisation of the framework should not oblige us to use any specific methods or
techniques; instead, it suggests what kinds of techniques can be used in rigorous
implementation of requirements and how these techniques can be integrated. The
framework can be used as a basis for improving software development areas that
are related to requirements implementation.

A theory-testing research approach is applied to validate the usability of the

developed framework in an industrial environment. The validation includes two
case studies within embedded software development organisations. First, a
current state analysis of an organisation's requirements implementation practices
in software development process is carried out with an historical
experimentation research approach according to Zelkowitz and Wallace (1998).
After the analysis, an improvement proposal is composed based on a comparison
between the current state and the developed framework. Finally, the proposal is
evaluated by the case organisation to get feedback on the applicability of the
framework.

11
 1.3 Scope

This research concerns both requirements engineering and software

implementation. Requirements engineering gives input to software
implementation and pervades its management elements over the whole software
development life cycle. On the other hand, there is a significant flow of
information from software implementation to requirements development
activities. For example, requirements are very hard to know up-front, and
therefore they tend to refine as the software matures. In this research, related
areas of requirements engineering and software implementation are covered and
their interrelations are studied.

The research is not merely concerned about methods that elaborate requirements
into software analysis models, designs, source code and test cases. Neither does
the research concentrate purely on the requirements management issues. Instead,
the scope of the research is to study rigorous implementation of requirements,
consistency management between requirements and software work products, and
how requirements and change management can be integrated into software
development activities.

The research is restricted to study requirements implementation in embedded

software development context. Many of this research's concepts, however, are
applicable for any kind of software development. The requirements
implementation framework's usefulness is validated in embedded software
environment with industrial case studies by making a current state analysis and
suggesting improvement proposals. However, implementing the proposed
improvements and then evaluating the possible improvement of the software
process is not within the scope of this research.

1.4 Structure

The structure of this research is illustrated in Figure 1. Chapter 2 introduces the

basic concepts of the study. Special characteristics of embedded systems and
embedded software development are discussed. Furthermore, definitions for
requirements and requirements engineering are given.

12
Chapter 3 examines the relationship between requirements software
development activities such as requirements analysis, design, coding, and
testing. Also elements needed for effective implementation of requirements in
the development activities are presented.

Embedded systems Requirements in Requirements

and requirements software development implementation-
(Chapter 2) activities supporting elements
(Chapter 3) (Chapter 4)

Conclusions Case study Requirements

(Chapter 7) (Chapter 6) implementation
framework
(Chapter 5)

Figure 1. Structure of the research

Chapter 4 presents requirements implementation-supporting elements which are

used throughout the software development. These elements include change
management, traceability of requirements, and consistency management between
requirements and different software work products. The first and second
research sub-questions are answered in Chapters 3 and 4.

Chapter 5 answers the third research sub-question by presenting a framework

that integrates elements of effective requirements implementation discussed
previously in Chapters 3 and 4. The sub-question is further handled in Chapter 6,
where the validation of the framework is discussed. Two case studies that
validate the framework are presented and the most important findings are
discussed.

The last chapter concludes the research and discusses the results. Further study
opportunities based on this research are also given. Appendix 1 contains a
questionnaire that can be used to clarify an organisation's current state of
requirements implementation practices. A summary of requirements
implementation practices and methods referenced to in this study is given in
Appendix 2.

13
 2. Embedded systems and requirements
This chapter introduces the basic concepts of the research. Because the focus of
this research is requirements implementation in embedded software domain,
special characteristics of embedded systems, embedded systems development, and
embedded software development have to be taken into account. Furthermore,
definitions for requirements and requirements engineering are given.

2.1 Embedded systems

Embedded systems are electro-mechanical products which relate mechanics,

hardware and software. Examples of embedded systems are mobile phones,
medical instruments and petrol dispensers. (Taramaa et al. 1998.) An embedded
system contains a computer which is part of a larger system that provides non-
computing features to the user (Douglass 2000). The size of an embedded
system might vary from a small thermometer to a chemical plant's process
controller.

Generally, an embedded system is composed of components implementing different

hardware and software functions. Hardware functions may be implemented by off-
the-shelf components, programmable logic and Application-Specific Integrated
Circuits (ASICs). Software functions may be implemented by off-the-shelf
processors such as Digital Signal Processors (DSPs) and other specific processors.
(López et al. 1998.) Embedded systems typically contain a small microcontroller
and limited software. However, with the development of such systems more
sophisticated microcontrollers, DSP chips, off-the-shelf real-time operating systems,
and software are being used. (Stankovic 1996; López et al. 1998.)

Embedded systems that have timing constraints are called real-time systems,
which not only have to produce correct output, but also should produce it at the
right time. Embedded real-time systems often operate in tight timing constraints,
where missing an important deadline can lead to severe consequences.
(Stankovic 1996.) Some real-time embedded systems operate in a safety-critical
domain, where functionality and reliability are critically important features.
Examples of safety-critical real-time embedded systems are avionics, traffic
control systems, and medical devices.

14
 2.2 Embedded systems development

The development of embedded systems is difficult, because these systems are

part of a physical environment whose complex dynamics and timing
requirements they must pursue. Taramaa et al. (1998, p. 907) define other key
characteristics of embedded systems development:

• embedded systems design is often constrained by the implementation

technology
• different technologies used in the same system can be developed
simultaneously
• the system can possess both low-level and high-level software
implementation
• maintainability and extendibility of embedded systems is achieved with new
functions, technologies, and interfaces
• short development time

Although development of embedded systems has been a common practice since

the first microprocessors, traditional development approaches are no longer valid
to ensure high quality under strict time-to-market and design cost constraints.
One of the reasons for this is the increasing complexity of today's
microprocessors and associated real time software. Also the complexity of
hardware and software interface communication mechanisms increases the need
for more suitable development approaches. (López et al. 1998.)

As a solution, more sophisticated development approaches for such complex

embedded systems are being developed. These methodologies are called
hardware/software co-design methodologies. (López et al. 1998.) The traditional
sequential development approach has been replaced by concurrent development
of hardware and software in the co-design approach. Hardware and software
design activities may begin with immature system specifications and
architectural designs. Concurrent development with incomplete specifications
requires close collaboration from all participants. For example, interfaces
between software and hardware must be designed in co-operation by both
hardware and software designers. Also testing processes for hardware and

15
software have to be integrated. Reusing components from previous designs or
outside the design group is considered as the main design goal. (Ernst 1998.)
Embedded systems co-design process is shown in Figure 2.

Requirements definition

Specification

System architecture

SW Development Interface design HW Design

SW HW
Designer Designer

Integration and test Reused

components

Figure 2. Embedded systems design process (adapted from Ernst 1998, 46).

The functionality of an embedded system is usually fixed and is mainly

determined by its interaction with the system's environment. An embedded
system often has numerous operation modes, capability to respond to exceptions,
and features that demand concurrent execution of different operations. (Gajski &
Vahid 1995.) These functions are handled by the embedded system's built-in
software, which is closely connected to the hardware. The close connection
between software and hardware means that the software development is often
steered and restricted by the hardware. (Lee 2000.)

2.3 Embedded software development

Software development is one component of the whole embedded systems

development process. Embedded systems development from the viewpoint of

16
 software is presented as the classical V-model according to Easterbrook (2001)
and the terminology of ISO/IEC 12207 in Figure 3. This model's concepts are
used throughout this research. However, system level analysis, design, and
testing activities are not in the scope of the research.

System System
Requirements Testing
Analysis

System System
Architectural Integration
Design Testing

Software Software
Requirements System
Analysis Testing

Software Software
Architecture Integration
Design Testing

Software Software
Analyse and Detailed Module Integrate and
design Design Testing test

Software
Coding

Figure 3. Software development V-model (based on Easterbrook 2001 and

ISO/IEC 12207).

The V-model basically states that the system requirements on the left side of the
V are gradually refined in the analysis and design activities until they are
implemented as software code. The activities on the right side of the V then
integrate and test the software until it is ready to be integrated with other

17
possible systems and hardware. The dotted lines between the development
activities indicate that the testing activities are based on the analysis and design
activities. Although verification and validation of the product is closely related
to testing, all the work products from the development activities must be verified
and validated throughout the development. Verification means checking that an
activity's work products fulfil the requirements or conditions imposed on them in
the previous activities. Validation ensures that the requirements and the final
product fulfil the product's specific intended use. (ISO/IEC 12207.)

It should be understood that the V-model is a simplified representation of

software development. There is a constant information flow from lower-level
development activities such as coding and design to higher-level requirements
analysis and design activities. The development activities are not separate
entities performed by different people; on the contrary, there is a lot of overlap
and collaboration among these activities and system and hardware development
activities. Furthermore, the software development activities interrelations and
performance order may vary.

Programmers write today's embedded software using low-level programming

languages, such as C or even assembly language, to cope with constraints on
performance and cost. Object-oriented languages such as C++ are not as
popular. (MOOSE 2002b.) Implementation and debugging tools for embedded
software are basically the same as those for traditional software, such as
compilers, assemblers, and debuggers. However, most tools for embedded
software development are immature compared to traditional software
development tools. (Sangiovanni-Vincentelli & Martin 2001.)

2.4 Requirements

A requirement is something that the system must exhibit or a quality that the
system must have. Requirements are defined at the beginning of the
development as a specification of what should be developed. Requirements for a
system are typically a combination of demands from different people at different
levels of the developing organisation and from the environment where the
system must operate (Sawyer & Kotonya 2001, 2). In addition to the description
of what the system should do, requirements often involuntarily describe how the

18
system should be done. Therefore, requirements are a mixture of problem
information, statement of behaviour, and design and manufacturing constraints.
(Sommerville & Sawyer 1997.)

Requirements used in the development of embedded systems are generally

classified as system, hardware, and software requirements. This research
concerns itself mostly with system and software requirements. System
requirements are derived from various high-level requirements stated by
different stakeholders. These stakeholders are for example acquirers, end-users,
sales and marketing people, regulators, system developers, and other existing
systems. System requirements are expressed in the terms of the stakeholders'
domain and are usually documented in a system requirements specification
(Sawyer & Kotonya 2001).

Software requirements are detailed requirements for software derived from the
system requirements. Software requirements fall into two general categories:
functional and non-functional requirements. Functional requirements define
capability and behaviour of the system. A sample functional requirement would
be a system's capability to format some text or modulate a signal. (Sawyer &
Kotonya 2001.) Non-functional requirements refer to performance, interface,
quality, and design constraints of the software. For example, a non-functional
requirement could define a recovery time for the software or specify the
programming language used in the implementation of the software. Software
requirements categorisation based on IEEE 830-1998 standard with explanatory
questions is listed in Table 1.

19
 Table 1. Categorisation of software requirements.
Functional What is the software supposed to do?
requirements
Performance What are availability, response time, and recovery time of
requirements software functions?
External interface How does the software interact with external entities, such
requirements as hardware?
Quality attributes What are the quality considerations of software such as
reliability, usability, maintainability, and portability?
Design constraints Are there any required standards in effect, implementation
languages, operating environments, etc.?
Non-functional requirements
Functional requirements

In order to achieve quality in the developed product, it is not enough only to fulfil
the functional requirements. When a non-functional requirement such as usability
is neglected, the resulting system might work correctly but it may be difficult to
use. Unfortunately, non-functional requirements are often not as easily verifiable
as functional requirements during the development. (Ebert 1997.)

Functional and non-functional requirements for software are typically

documented in a software requirements specification (SRS). The readers of an
SRS are expected to have knowledge of software engineering concepts, which
can reflect on the language and notation used in the specification. (Sawyer &
Kotonya 2001.) In addition to the software requirements that are stated in natural
language, graphical representations are often needed to depict a full picture of
the intended system. These representations as analysis models also help
developers to detect inconsistencies, errors, and omissions in the requirements.
Furthermore, graphical descriptions improve developers understanding of the
requirements. (Wiegers 1999.) The SRS and its graphical representations are
discussed further in the next chapter.

Requirements for embedded software have a different emphasis than

requirements for conventional desktop software. In conventional software
development, non-functional requirements such as timing, reliability, robustness,
and power consumption are secondary to the logical correctness of
computations. In embedded software development, non-functional requirements

20
are equally important as functional requirements. (Karsai et al. 2003.) Non-
functional requirements are especially recognised as an important success factor
in market-driven embedded software development (Punter et al. 2002). A
product with outstanding functionality is not enough; instead, both functionality
and quality characteristics are needed for a successful product.

In addition, system and hardware requirements are more important in embedded

software development than in development of conventional software. During the
system design, an agreement based on system requirements is made about the
system's components and their purposes. Then, software and hardware
requirements are elaborated for each of these components separately. (Davis
1990.) Therefore, development of embedded software is dependent on system
and hardware requirements. Developers of conventional software hardly need to
worry about system level requirements (except for large software systems) and
even less so about hardware requirements.

2.5 Requirements engineering

The term requirements engineering has various definitions. Robinson and

Pawlowski (1999) characterise requirements engineering as an iterative process
of discovery and analysis of agreed set of clear, complete, and consistent system
requirements. This research shares a broader perspective on requirements
engineering, including the management viewpoint, as stated by Dorfman (1997,
p. 12): "...requirements engineering consists of elicitation, analysis,
specification, validation/verification, and management [of requirements]."

Requirements engineering can be further classified into system and software

requirements engineering. System requirements engineering is concerned with
analysing and documenting system requirements. Requirements engineering
from systems engineering viewpoint is shown in Figure 4. In system
requirements engineering, stakeholders' needs are transformed into system
description, system performance parameters, and system configuration. System-
level requirements engineering is also concerned with the partitioning of the
system. This is achieved by identifying which requirements should be allocated
to which components (e.g. software, hardware, or subsystem). (Sawyer &
Kotonya 2001.) The allocation is based on different criteria such as cost of

21
implementation, cost of replication, implementation time, ease of change, and
error rates during operation (Stevens et al. 1998).

Stakeholders'
needs and constraints

System requirements
development

Allocation
Flow-down

Detailed sub-system Detailed sub-system

requirements requirements
Software Software
requirements requirements
Hardware Hardware
requirements requirements

Figure 4. Requirements engineering in systems engineering context (adapted

from Parviainen et al. 2003, 13).

After the allocation, a flow-down process will take place, where requirements
for the lower-level elements are specified in response to the allocation. The
lower-level elements must have at least one requirement that responds to the
allocation, but often many requirements are written for one allocation. (Dorfman
1997.) Finally, requirements are implemented in sub-systems either in hardware
or software disciplines.

Software requirements engineering is concerned with establishing and

documenting software requirements. Allocated system requirements are
transformed into software requirements through the use of analysis, design,

22
trade-off studies, and prototyping. The major difference between system and
software requirements engineering is that the origins of system requirements are
stakeholders' needs, while the origins of software requirements are system
requirements. (Thayer & Dorfman 1997.) However, in small scale development,
software requirements may come directly from stakeholders.

Sawyer and Kotonya (2001) have defined requirements engineering activities,

which are broadly compatible with the ISO/IEC 12207 (1995) software life cycle
processes standard. Requirements engineering consists of activities such as
elicitation, analysis, specification, validation, management of requirements.
These activities, except for requirements management, should comprise an
iterative process where activities are repeated until an acceptable requirements
document is produced. This process is not a separate front-end activity in the
software life cycle, but rather an activity that is initiated at the beginning of the
life cycle and continues to be refined through the life cycle. Requirements
management is an activity that spans the entire software development. It consists
of requirements identification, tracing, and change management. (Sawyer and
Kotonya 2001.)

23
 3. Requirements in software development
activities
After the agreed system requirements have been specified and further allocated
to software, the work begins on evolving these requirements into a software
requirements specification, designs, code, and test cases. This chapter discusses
the relationships between requirements and software development activities of
the V-model presented in the previous chapter. In addition, elements needed for
effective implementation of requirements in the development activities are
presented. These elements were gathered from current literature, which includes
software engineering standards, research papers, and books. The presented
elements are later used in Chapter 5 to construct the requirements
implementation framework.

3.1 Software requirements analysis

Software requirements analysis is a software engineering activity that brings

system requirements closer to software designs. In this activity, system
requirements are elaborated into software requirements. The elaboration is done
by refining, analysing, modelling, and specifying allocated system requirements
into software requirements. To increase knowledge about what is required from
the software, the problem is partitioned and representations that describe
requirements for software are constructed. Thus, requirements analysis is a
technical step, where broad expressions of the software are refined into a more
detailed specification that serves as a foundation for all other software
engineering activities that follow. (Pressman 2000.)

Two distinct activities are performed during the specification of software

requirements: problem analysis and product description. Problem analysis is an
activity where analysts brainstorm ideas, interview people, and identify all
possible constraints on the problem's solution. Product description activity is
concerned with the external behaviour of the product. The purpose of the
product description activity is to solve the problem that is understood after the
problem analysis. (Davis 1990.) Possible approaches for these activities are
structured analysis, object-oriented analysis, and formal methods. Sample

24
techniques include data flow diagrams, state-transition diagrams, and use cases.
(Parviainen et al. 2003.)

A software requirements specification (SRS) is developed as a result of the

requirements analysis. If necessary, it may be jointly prepared with customers,
because software engineers usually do not understand customers' problems and
domain area well enough to produce correct requirements (IEEE Std 830-1998).
On the other hand, software developers who will later implement and test these
requirements should be also included in the software requirements analysis
activity. For example, software architects could prevent the acceptance of
unachievable requirements. Benchmarking and prototyping the architecture
before the requirements are nailed down may reveal problems with the suggested
requirements. (Mead 1994.) Software testers should be included in requirements
reviews to make sure that the requirements can be eventually verified. All the
requirements included in the SRS should have a proper verification criterion.
(Wiegers 1999.)

Requirements in an SRS should be documented with related attributes so that

they are identified and prioritised. Identification is needed for requirements
management. Priorities for requirements should be assigned to define
implementation order and allow resolution of conflicts in later development
activities. Customers naturally demand high priorities for requirements that are
important for users. In contrast, developers desire high priorities for
requirements that are important for software development and maintenance. The
cost of a certain requirement plays also major role in deciding what is essential
to implement. Reconciliation of these different viewpoints can be supported, for
example, with Quality Function Deployment (QFD) and Wiegers' method.
(Wiegers 1999.)

A requirement's stability may be also be stated as an attribute. For some

requirements it is possible to define the probability of their change in the future.
The stability attribute enables software developers to construct designs that are
more tolerant of change. Other possible attributes for requirements include
source and scope of the requirement. (Sawer & Kotonya 2001.)

An SRS must be in line with system requirements. To achieve this, traceability

between system and software requirements must be established. Furthermore,

25
software requirements consistency with system requirements must be ensured.
(ISO/IEC 12207:1995, IEEE 830-1998.) The SRS should be comprehensive and
should not include any design, implementation, testing, or project management
details other than applicable design and implementation constraints. An
undocumented requirement is not part of the agreement and no one should
expect it to be in the product. Although all the needed requirements should be
contained in the SRS, software design and construction can begin before the
entire SRS is written. This requires that the development project has baselined
agreement for a set of requirements. Baselined requirements are approved and
reviewed requirements from the SRS under development. Changes to the
baselined requirements should be made through a defined change control
process. (Wiegers 1999.)

After the software requirements have been analysed and specified, they need to
be verified. Verification ensures that requirements (1) correctly describe
behaviour and characteristics of the system, (2) are internally consistent, (3) are
complete and high-quality, and (4) provide an adequate basis to proceed with
design, implementation, and testing. Verification is not a solitary activity
performed after all of the requirements have been documented. Instead, as the
requirements tend to develop during the requirements gathering and analysis,
they need to be verified incrementally. (Wiegers 1999.)

One possible approach for verification of software requirements is a formal

inspection of the SRS. This is especially useful when requirements are being
baselined for further development of software. Reading techniques such as
perspective-based reading (Shull et al. 2000) can be used in requirements
inspections. In addition to inspections, some quality attributes, such as safety,
might require their own specific verification procedures (Rodríguez-Dapena et
al. 2001).

3.2 Requirements and software architecture design

Software's architectural design depicts a holistic view of the software to be built.

It represents the structure and organisation of software components, their
properties, and the connections between them. The architecture enables software
engineers to analyse the effectiveness of the design in meeting its requirements.

26
Furthermore, risks associated with the construction of the software are reduced
by comparing alternative architectural solutions. (Pressman 2000.) Having a
good software architecture does not ensure that the developed product will meet
its requirements. On the other hand, having a badly designed architecture makes
it almost impossible to meet the product's requirements. (Hofmeister et al. 2000.)

Requirements are a key input to the design of software architecture. Elaborating

requirements into valid software architecture that satisfies those requirements is
a difficult task and is mainly based on intuition. Little is known about the impact
of architectural choices on the requirements. (Grünbacher et al. 2001.) The
architectural design must meet both functional and non-functional requirements.
For example, performance requirements and timing constraints for embedded
software influence architectural design. Sample constraints include data
throughput rates, function call latencies, and interrupt congestion (Ronkainen et
al. 2002).

Other quality attributes such as testability, understandability, and modifiability

can be also implemented in the architectural design. This is achieved, for
example, by specifying required levels of cohesion and coupling for the design
(Davis 1990). Some of these quality attributes, such as safety, may need specific
verification processes to ensure that these characteristics are included in the
design (Rodríguez-Dapena et al 2001). Ebert (1997) suggests to use checklists-
based inspections for measuring the success of quality attributes absorption into
the architectural design.

An important purpose of the software architecture is to abstract the software

design so that architects can analyse trade-offs between different solutions.
Trade-off analysis means exploring the impact of certain decisions in terms of
all the criteria and constraints (Ruhe et al. 2003). Based on a trade-off analysis,
the architect must decide which requirements gets higher priority, and design a
solution that adequately satisfies the requirements. A requirement might be
sacrificed or compromised as a result of the trade-off analysis. (Hofmeister et al.
2000.) Sample approaches for analysing architectural designs against
requirements include Architecture Trade-off Analysis Method (ATAM),
Scenario-Based Architecture Reengineering (SBAR), and Architecture Quality
Analysis (AQA) (MOOSE 2002a). In addition to trade-off analyses, these

27
methods can be used for evaluating absorption of quality attributes into the
design.

As architects examine the requirements and develop possible architectural

solutions, they might need to consult stakeholders to negotiate on changes to the
requirements (Hofmeister et al. 2000). Ideally, software architects are involved
in the early activities of the development. This way development of
unachievable requirements is more likely to be prevented. Software's
architecture must be considered during the development of requirements to make
certain that the requirements are valid, consistent, and feasible. Architecture
modelling and trade-off analysis are important activities and should already take
place during requirements engineering. (Mead 1994.)

Software projects often consider requirements analysis and candidate

architectural solutions concurrently. However, the relationship between
requirements and software architecture is not obvious. Requirements describe
aspects of the problem to be solved and constraints on the solution. In contrast,
architecture describes a solution to the problem stated in the requirements.
Requirements deal with such concepts as goals, options, agreements, issues,
conditions, and system features and properties. They may be simple or complex,
precise or ambiguous, stated in natural language or precise formalism. In
contrast, the concepts used in architectural design differ from those used for
requirements. Architecture deals with components, connectors, topologies, and
the software's desired properties. These differences between requirements and
architectures make it difficult to build a bridge that spans the two. (Grünbacher
et al. 2001.)

Because of the different concepts used in requirements and software

architectures, consistency and traceability maintenance between them is
complicated. A single requirement may concern multiple architectural solutions
and a single architectural element may have numerous relations to different
requirements (Grünbacher et al. 2001). Nevertheless, software development
standards require that traceability between an architectural design and
requirements is established. Furthermore, the architecture's consistency with the
requirements must be ensured. (ISO/IEC 12207; ISO/IEC TR 15504-5.)

28
 3.3 Requirements and detailed software design

Detailed software design represents the software at a level of abstraction that is

very close to code. The design must describe data structures, interfaces, and
algorithms in sufficient detail to guide developers in construction of the software
code. These descriptions are represented using design notations such as
flowcharts, N-S charts, decision tables, program design languages (PDL),
dataflow diagrams, entity-relationship diagrams and UML. (Pressman 2000.)
Object-oriented design of embedded real-time software considers detailed design
as definition of data structures, associations, operations, algorithms and
exceptions (Douglass 2000).

Software designers should be included in the development of requirements (e.g.

in requirements inspections) to make sure that requirements can serve as a basis
for detailed designs. Several different software designs will usually satisfy the
given requirements. These designs may vary in performance, efficiency, and
robustness characteristics. The most efficient way to fulfil the requirements
should be explored. (Wiegers 1999.)

Both functional and non-functional requirements affect the detailed design

activity. While functional requirements steer the design's operations and
algorithms, non-functional requirements set constraints on these solutions. For
example, performance requirements oblige designers to develop algorithmic
solutions that meet the given timing constraints (Davis 1990). Also hardware
constraints, such as available memory, affect designers' work by restricting the
solution space. Design constraints stated in software requirements affect the
detailed design activity explicitly. These constraints may define the notations
and techniques used in the detailed design.

Quality attributes appear in design trade-offs when designers need to choose

between particular structural and behavioural aspects of the system. For example,
quality attributes can be used to guide the selection and application of design
patterns during design. Characterising and using patterns according to quality
attributes allows designers to meet system-wide non-functional requirements.
(Gross & Yu 2001.) As in architectural design, some quality attributes may need
specific verification processes to ensure that these characteristics are reflected in
the detailed design (Rodríguez-Dapena et al 2001).

29
Software development standards require that a software component's detailed
design is traceable to software requirements and consistent with them (ISO/IEC
12207:1995; ISO/IEC TR 15504-5:1998). The design should be also verified so
that all of the functional requirements are accommodated and unnecessary
functionality is not included (Wiegers 1999).

3.4 Requirements and software coding

The purpose of the software coding activity is to produce software units that
reflect software designs and requirements. In coding activity, the developers
begin to write modules that compile in the programming language (Mazza et al.
1996). If designs are specified detailed enough, they may be used directly as the
basis for coding. Otherwise, developers need to use software requirements stated
in the SRS and detailed designs together.

Implementation of functional requirements in code is more straightforward than

is the case with non-functional requirements. Software engineers have to
consider various aspects while implementing non-functional requirements.
Performance requirements might oblige software engineers to use a low-level
programming language (Ronkainen et al. 2002). Efficiency requirements, such
as memory constraints, require programmers to use available hardware resources
narrowly (Davis 1990).

Maintainability and portability requirements are also clearly present in the

coding activity. The source code's maintainability is achieved by following the
organisation's coding standards. Static code analysis tools such as QAC/++
(2003) or Logiscope (2003) can be used to identify a source code that is not
conformant to the specified quality attributes. These tools typically identify
problem areas in the code's structure, maintenance, and portability issues.
Programming language selection, for example between a standardised high-level
and a processor specific low-level language, also has considerable influence on
portability requirements.

Certain quality attributes, for example predictability, should become crucial for
developers at the coding activity. Definition of these characteristics cannot be
made precisely at the beginning of the development because of their dependency

30
on the software architecture and implementation technology. (Rodríguez-Dapena
et al. 2001.) On the other hand, trade-off analyses between some quality
attributes have to be made before coding. For example, fulfilling an efficiency
requirement may require that a code is implemented using a specific compiler
and operating system. As a result, the system may work fast, but at the same
time it will be hard to maintain, enhance, and port to another environment.
(Wiegers 1999.) Some of these quality attributes may need specific verification
procedures to ensure that the desired quality is elaborated into software code
(Rodríguez-Dapena et al. 2001).

In some phase of the construction process, developers will encounter ambiguity

and confusion while translating software requirements into code. To solve these
problems, the obscure requirements should be traced back to their sources and a
solution would be negotiated. Procedures should be defined for a situation where
problems can't be resolved immediately. (Wiegers 1999.) Management of the
confusion and ambiguity is improved when requirements are traced into code
and their consistency is checked. Maintaining a code's traceability and
consistency with requirements is also required by software development
standards (ISO/IEC 12207:1995; ISO/IEC 15504:1998).

3.5 Requirements and software testing

The purpose of the software testing activity is to ensure that the produced
software will satisfy the software requirements. Software testing includes three
basic testing stages: module, integration, and system testing. Module testing
verifies that a software module's source code is doing what it is supposed to do.
Testing of software modules is followed by integration testing, where the focus
is on software designs and the structure of the software architecture. Finally,
system testing is performed to ensure that the software as an entity fulfils all
software requirements. (Mazza et al. 1996; Davis 1988; Pressman 2000.)

System testing is the most significant testing activity from software requirements
viewpoint. Davis (1988) emphasises that system testing should ensure that the
entire software embedded in its actual hardware environment behaves according
to the SRS. System test plans and test cases should be designed to ensure that all
the software's desired characteristics are achieved. The purpose of test planning

31
is to assess how the software will be tested for conformity with the SRS.
Another purpose of the planning is to ensure that the SRS is verifiable. An SRS
is verifiable only when all of the stated requirements are verifiable. However,
some of the requirements cannot always be rigorously verified. For example,
usability requirements have multiple interpretations and are hard to measure
quantitatively. (Davis 1990.)

Software testing has a major role in embedded system development due to the
system's reliability and other non-functional requirements. For example, a
system's quality attributes such as safety and reliability often require high code
coverage percentages. Usability aspects could be tested by gathering suitable
information during usability tests in an environment as close to the operational
environment as possible (Ebert 1997). Other sample techniques for non-
functional requirements testing include recovery testing, security testing, stress
testing, and performance testing (Pressman 2000). However, the realisation of
quality attributes is often difficult to verify. Various possible scenarios and
different input combinations force developers to limit the set of test cases. In
reality, quality attributes are often the first ones to bypass systematic
verification. (Rodríguez-Dapena et al. 2001.)

The software testing activity's test cases are based on software requirements. The
relation between requirements and test cases is considered so essential that the
link between them should be always established (Weber & Weisbrod 2002).
Every requirement should be traced to at least one test case, so that all expected
system behaviour is verified (Wiegers 1999). Composing test cases as soon as
requirements stabilise enables developers to find and correct flaws in
requirements inexpensively. Writing test cases for requirements crystallises the
developers' vision of how the system should behave. Ambiguous and conflicting
requirements should be revealed when the developer cannot describe the
system's expected response to a test case. (Wiegers 1999.)

In actual practice, software developers often fail to manage requirements,

validation criteria, and test cases. Although it is agreed that requirements
specifications should go hand in hand with validation criteria and test cases, the
actual specification documents do not really fulfil these agreements. One reason
for this is the fact that it is very difficult to specify validation criteria and suitable
test cases for some functional requirements. With non-functional requirements the

32
case is even harder. Providing concrete requirements validation criteria examples
for developers and having developers exploit and re-use these samples may
improve these problems. The validation of non-functional requirements is also
improved if these requirements are stepwise refined until they are implemented by
a set of functional requirements. (Weber & Weisbrod 2002.)

33
 4. Requirements implementation-
supporting elements
The previous chapter discussed the role of requirements in software
development activities and overviewed elements needed for effective
requirements implementation in these activities. The importance of maintaining
consistency and traceability between requirements and software work products
was emphasised in all discussed activities. In addition, requirements change
management is needed for controlling the inevitable change.

This chapter discusses these requirements implementation-supporting elements,

which manifest themselves throughout the software development. First,
requirements change management and its effect on software development is
discussed. Then, traceability of requirements, which is also an important part of
the change management, is covered in more detail. Finally, approaches for
ensuring consistency between requirements and different software work
products are presented. The presented requirements implementation-supporting
elements are later used in Chapter 5 to develop the requirements implementation
framework.

4.1 Requirements change management

Changing requirements are the reality in system development rather than stable
ones. Designers are faced with unpredicted and disruptive changes in
requirements that the system has to satisfy. (Harker et al. 1993.) Requirements
sometimes change because of an error in requirements analysis, but more often it
is consequence of a change in the system's environment (Sawyer & Kotonya
2001). In addition to the change in the system's operating environment, the
hardware where the software is embedded may change. Therefore, concurrent
development of the embedded system's hardware and software is a possible
source for change. (Mäkäräinen 2000.) Sometimes, unfeasibility of requirements
is only revealed once the actual software designs and code are implemented and
tested; therefore, original requirements may need to be changed because of a
change request from software development activities.

34
Whatever the reason for a change, it is important to manage the change by
ensuring that the proposed changes go through appropriate review and approval
procedures (Sawyer & Kotonya 2001). Effective system development requires
organisational change management policies, which define the processes used for
change management and the related information that is associated with change
requests (Kotonya & Sommerville 1998). The need for requirements change
management is also stated in the Software Engineering Institute's process
improvement framework Capability Maturity Model Integration (CMMI). The
requirements management process area in the CMMI framework requires that
changes to the requirements are managed during the project (CMMI-SW 2002).

The need for a formal requirements change process arises when many changes to
requirements are proposed. A requirements change management process consists
of activities for documenting, reporting, analysing, costing, and implementing
changes to requirements. Change management process can be thought of as a
three-stage process. First, the problematic requirement is identified and analysed
using the problem information and change for the requirement is proposed.
Then, the proposed change is analysed to see how many other requirements are
affected and what is the cost of the change. Finally, the change is implemented
and relevant documents are updated. (Kotonya & Sommerville 1998.)

Procedures for requirement problem analysis and change implementation are

dependent on the type of the change, the requirement, and requirements
documentation. Cost and change impact analysis, however, are more general
processes. These processes ensure that unnecessary changes are not made and
that affected requirements are discovered. (Kotonya & Sommerville 1998.)

A requirements change management process may be included in a process that is

intended for general change management in the system development. Examples
of change management models include Olsen's change management model, the
spiral-like change management model, and the generic change management
process model (Parviainen et al. 2003).

A recent survey in embedded software industry revealed that there are problems
implementing requirements change management process. Although change
management procedures are defined, they are in many cases rejected. The main
reason for this is that these procedures take too much time. Time is consumed

35
for example in re-reviewing all the relevant documents after every change. As a
result of inappropriate change management, requirements and design documents
may easily lose their consistency. (MOOSE 2002b.)

Change management process is also complicated by communication problems

specific to embedded systems development. Concurrent development of an
embedded system is often carried out in separate technology groups. Therefore,
detailing the changes to all concerned parties becomes crucial when developers
share the same software components. However, the development groups usually
operate independently and their working practices, tools, and vocabularies are
not always similar. (Mäkäräinen 2000.)

To overcome the problems in implementing a change, a change control board or

(CCB) may be needed for controlling the change process. Such board is a body
of people (possibly one individual or a group of people) that makes decisions
about approving proposed requirement changes. Generally, a CCB reviews and
approves changes to any baselined work product on the project. Large projects
often have many levels of control boards. For example, some control boards are
responsible for business decisions while others deal with technical issues.
Smaller projects may have only one or two people that make all the change
decisions. However, a CCB should have representatives from all groups that are
affected by a change. Sample representatives include people from product
management, project management, configuration management, software
development, and quality assurance. (Wiegers 1999.)

4.1.1 Impact analysis

The impact of the change must be analysed before a requirement change is

approved. Not only must the change's impact on the requirement specification be
analysed; its impact on lower level work products has to be examined as well.
For example, if a change is made to a single requirement that affects few details
in high-level designs, it may affect many other lower-level designs,
implementations and test cases. This kind of phenomenon is know as the ripple
effect. The purpose of impact analysis is to tackle problems such as the ripple
effect by identifying potential consequences of a change. Impact analysis can

36
also be used to determine what to modify in order to accomplish a change.
(Arnold & Bohner 1993.)

Research on impact analysis has concentrated mainly on determining source

code level changes or using traceability to predict changes to different work
products created during the development. Techniques for impact analysis on
implementation level include control and data dependency, and program slicing.
Various traceability approaches enable impact analysis throughout the software
development. (O'Neal & Carver 2001.) Wiegers (1999) suggests that developers
use checklists and defined procedures for discovering possible implications of a
change. As a result of impact analysis, a developer reports to the CCB the
needed effort for a requirement change. A standard reporting template will make
it easier for the CCB to find information needed to make decisions.

Quality impact analysis methods such as Quality Factor Deployment can be

utilised to assess the impact that a requirement change is likely to have on
quality aspects such as performance, safety, and reliability. Also requirements
inspections, viewpoint analysis and trade-off analysis can used to identify
potential conflicts and trade-offs within the set of changing requirements. (Lam
& Shankararaman 1999.)

In actual practice, impact analysis is mostly performed manually, because tool

support is weak. Also absence of traceability information between requirements
and other software work products complicates impact analysis. (MOOSE 2002b.)
Wiegers (1999) states that the organisation's ability to succeed in impact analysis
depends on the quality and completeness of the maintained traceability data.

4.1.2 Change management tools

When many requirements changes are introduced, the need to use a change
management tool grows. Requirements change management may be supported
by specialised requirements management and software configuration
management tools. Some of these tools enable the use of database supported
electronic change request forms. Furthermore, they may provide automatic
notification to responsible persons with electronic forms. (Kotonya &
Sommerville 1998.) Sample commercial requirements management tools that

37
also support requirements change management include Rational RequisitePro
(2003) and DOORS (2003).

The general problem with change management tools is their own implicit model
of change process, which organisations must adopt. In addition, special-purpose
change support tools are fairly expensive and difficult to integrate. Therefore,
these tools are mostly used by very large organisations in very large projects.
Because requirements management also involves release management and
configuration management, a tool support that integrates these features is
actually demanded by industry. Some of the requirements management tools
provide only some of the demanded features, while integration of configuration
management, change management, and test case management is needed.
(MOOSE 2002b.)

Requirements change management and software configuration management

(SCM) have similar objectives. SCM is used as a means of controlling change
throughout the software development process. It consists of activities that identify
unstable work products, establish relations among them, define mechanisms for
managing versions of these work products, control changes that are imposed, and
audit and report about the changes that are made. Specifically, change control
activity of SCM is responsible for ensuring quality and consistency as changes are
made to configuration items. (Pressman 2000.) Requirements can be regarded as
software configuration items and therefore controlled under SCM. The same SCM
processes and tools that are used for design, implementation, and testing activities
can be also used for versioning and managing changes in the requirements.
(Crnkovic et al. 1999.) However, Kandt (2002) states that most configuration
management tools are file based and therefore do not manage documents at a level
of granularity that is truly useful - individual requirements, design concepts,
classes, functions, test cases, and so on.

4.2 Requirements tracing

Requirements, designs, and implementations that are correct, consistent, and

error-free ensure that the developed system meets its stakeholders needs. One of
the key factors in achieving these features is understanding and tracing the
relationships amongst these work products. (Palmer 1997.) The development of

38
embedded systems also requires that traceability between a system's different
technology components is handled (Mäkäräinen 2000). Requirements
traceability is defined as "...the ability to describe and follow the life of a
requirement, in both a forwards and backwards direction (i.e., from its origins,
through its development and specification, to its subsequent deployment and use,
and through all periods of on-going refinement and iteration in any of these
phases)." (Gotel & Finkelstein 1994, p. 97)

Requirements traceability has various benefits in software development when

used appropriately. As discussed earlier, it provides a means of managing
change in requirements. Traceability can be also used for showing a system's
compliance with requirements, maintaining a system's design rationale, proofing
when a system is complete, and establishing maintenance mechanisms (Ramesh
et al. 1995). Traceability is considered a key issue in transforming requirements
into architecture and further work products (Grünbacher 2001). Design decisions
are consistent with decisions made earlier in system development when lower
level work products are traced to higher-level work products. Requirements
tracing enables conflict detection by discovering linkage between selected
entities and by providing visibility into the entire system. It also supports
assessment of high-level behaviour and non-functional requirements' impact on
design specifications. Also test cases coverage in relation to requirements is
provided. (Palmer 1997.) Tracing requirements to designs, software units and all
levels of testing has been also stated to be the best way to monitor progress
(Lindström 1993).

A defined traceability policy is needed as the number of requirements grows. All

decisions about which elements are to be traced and how to do so should be
defined in the project's inspection phase (Weber & Weisbrod 2002). Factors that
influence such decision-making are the number of requirements, the system
lifetime, organisational maturity, the development team size, type of the
developed system, and specific customer requirements. Small teams can handle
changing requirements without structured traceability information, but larger
teams have to establish a formal traceability policy. Certainly, the development
of critical systems needs a more comprehensive traceability policy than the
development of non-critical systems. (Kotonya & Sommerville 1998.)

39
Although requirements tracing is recognised as a very useful practice in software
system development, it is often misused and poorly performed (Gotel &
Finkelstein 1994; Lindström 1993; MOOSE 2002b). One of the main problems
with tracing is related to the developers' necessity to manually add trace
elements to requirements documents and following work products. Since these
products have little or no direct consequence on the work of the development
team, setting trace elements has a low priority. The benefits of traceability are
not understood until later activities of the development life cycle, such as
validation testing and system installation and operation. (Palmer 1997.) Another
reason for problems with tracing requirements is that relations amongst
documents (e.g. between requirements and architecture) are too complex to
specify manually especially when projects are large (MOOSE 2002b).

Development environments for embedded software rarely provide any support

for managing traceability issues (Mäkäräinen 2000). Common methods for
requirements tracing include traceability matrices in the form of lists and tables.
These matrices show relations among the requirements and relations between
requirements and development elements (e.g., design, code, or test case).
However, such methods require manual work and therefore have been regarded
as laborious. Especially, manual maintenance of traceability links has been
found to be difficult and error-prone (Mäkäräinen 2000). Tracing of quality
attributes is considered to be particularly complicated (Ebert 1997).

Tools for automating the process of tracing have been presented, but the problem
still remains in industry (MOOSE 2002b). Sample commercial requirements
tracing tools include RTM (2003) and DOORS (2003). Introducing a
requirements tracing tool is a big step for the company and often the tools do not
satisfy the needs of the company nor the traceability. Reasons for rejecting the
use of traceability tools are various. Requirements tracing tools have for
example been criticised for ignoring the meaning of relations between
requirements. The link between requirements is established but the rationale
behind the link is missing. (MOOSE 2002b.)

Although requirements traceability is regarded as problematic to implement in

practice, software development organisations should strive to utilise it. Software
standards require that software work products are traceable throughout the
development chain (ISO/IEC 12207:1995; ISO/IEC 15504:1998). The software

40
process improvement framework CMMI also emphasises the importance of
traceability. Requirements management process area in CMMI requires that bi-
directional traceability among the requirements and work products is maintained
(CMMI-SW 2002). However, tracing requires a great deal of resources and
therefore it should be a compromise reflecting costs and benefits of linkages.
Tracing should be executed only where the traceability information is truly
useful. (Stevens et al. 1998.)

4.3 Consistency management

Software work products at different levels of software development (e.g.

requirements, designs, code, and test cases) are closely inter-related. Maintaining
consistency among these work products is part of verification and validation
processes in software development. Consistency among the work products
ensures that the produced software will meet its specified requirements. Because
requirements serve as a starting point for software development, consistency
between requirements and following work products should be ensured
throughout development. Also software standards (ISO/IEC 12207:1995;
ISO/IEC 15504) and industrial experiences (Lindström 1993; Olsson & Runeson
2002) indicate that consistency between requirements and software work
products has to be maintained at all levels of development. The software process
improvement framework CMMI also states the need to handle inconsistency.
The framework's requirements management process area requires the handling
of inconsistencies between requirements and work products (CMMI-SW 2002).

Sources for inconsistencies between different software work products are various.
Software engineering is conducted by human beings and therefore errors may be
introduced in any development activity by the developers of the system.
Inconsistencies may also arise when an organisation's personnel and customers
change. As analysts and customers leave a development project, they are replaced
by others who might drive development in a different direction. As a result,
inconsistencies between new requirements and existing software work products
arise. (Robinson & Pawlowski 1999.) Other possible sources for inconsistency are
approaches and tools used in the software development. The problem with current
development approaches is that relating information across work products is either
not straightforward or not possible at all. In addition, most of the development

41
tools support only specific development activity with limited support to other
activities in the development process. (Olsson & Grundy 2002.)

Lindström (1993) reported typical problems in software development when

requirements consistency with lower work products is not closely monitored.
First, requirements were analysed and high-level designs based on these
requirements were derived at the beginning of the project. Then, detailed designs
were developed based on the high-level designs with little attention to the source
requirements. Similarly, source code and unit tests were derived from the
detailed designs, which removed the development further from the requirements.
Not until integration and testing activities did the focus return to the source
requirements. By this time, some requirements had changed, new requirements
had been added, and some requirements had been lost or forgotten. The result
was that inconsistencies between requirements and other software work products
produced an incorrect product. The lesson learned from this project was that
requirements compliance has to be somehow measured at each step of the
development. To achieve this, Lindström suggests tracing of requirements and
incorporating a requirements check in work product reviews at all levels of
development.

As the size and complexity of the software continues to grow during the
development, software developers are faced with problems in preserving
consistency between different work products. These problems are tackled with a
consistency management process, which is composed of various activities such
as the detection of overlaps, consistency checking, diagnosis, consistency
handling, consistency tracking, and specification and application of consistency
management policy (Kozlenkov & Zisman 2002).

Ensuring consistency between different work products and requirements has

been a research subject for some time now. The researchers have, for example,
proposed using inspections, quality models, and formal methods on proofing
consistency between requirements and following work products. Some of these
methods for design, coding, and testing activities are presented next.

42
Design
Despite the close relationship between requirements and software designs, little
attention has been paid to their integration. This has increased the risk of
inconsistencies in software development. A sample approach to reduce the risk
of inconsistency between requirements and architecture has been proposed by
Inverardi et al. (2001). They propose an approach that traces co-ordination
requirements from their definition to the low-level specification. The
architecture is then validated with respect to these co-ordination requirements.
Other sample methods to strengthen the relationship between requirements and
architectural design include CBSP (Component-Bus-System-Property) method
(Grünbacher et al. 2001) and the combined use of goal-oriented language GRL
and a scenarios-oriented architectural notation UCM (Liu & Yu 2001). Chechik
and Gannon (2001) present an approach where lightweight formal techniques are
used for automatic analysis of consistency between software requirements and
detailed design. Design inspections can also serve as a consistency checking
approach, as is shown by Travassos et al. (1999).

Coding
Approaches for ensuring consistency between requirements and software code
have also been presented. Robinson (2002) suggests an implementation
approach where an instrumented code is used to ensure that requirements are
traceable and satisfied. Chechik and Gannon (1995) propose a tool called
Analyzer which discovers instances of inconsistencies and incompleteness in
implementations by proofing state transitions. Punter et al. (2002) present a
method for evaluation and correct implementation of non-functional
requirements in software designs and code. They suggest combining the use of
quality modelling, probability concepts, and measurement techniques in order to
evaluate code and design artefacts. Phased inspection technique according to
Knight and Myers (1993) can be used for checking the existence of appropriate
functionality and quality characteristics such as portability, reusability, and
maintainability in code.

Testing
Testing can serve as a consistency checking approach throughout embedded
system development. Co-design of hardware/software systems relies heavily on
testing that is executed at all levels of design activities. When a consistent set of

43
test cases is used, consistency among different design levels can be ensured. To
support this, Gupta et al. (2001) propose an automatic test scenario generation
approach. Their method is based on a state-based model of the proposed system.
Another sample requirement-based approach for rigor verification of software
systems is suggested by Tahat et al. (2001), where Specification and Description
Language (SDL) is applied to create finite state machines, which are used to
automatically generate test cases that are consistent with the requirements.

44
5. Requirements implementation framework
In order to analyse and improve an organisation's capability to implement
requirements effectively, a framework is needed to describe how requirements
are ideally implemented throughout the development. This chapter presents such
a framework, which integrates elements of requirements implementation
referenced to previously in Chapters 3 and 4. These elements were found in the
literature as good practice of requirements implementation. While the previous
chapters provided examples of how these elements could actually be
implemented (summary available in Appendix 2), this chapter discusses their
interrelations and integration.

5.1 Requirements implementation framework for

embedded software

The purpose of the proposed requirements implementation framework is to

describe what kind of elements are needed for effective implementation of
requirements during embedded software development. These elements are, for
example, requirements implementation processes, methods, and roles. Because
elements such as a change control board, traceability, and extensive reviews
require considerable resources, smaller and more flexible organisations may
need to tailor the framework's concepts. Tailoring of the framework is discussed
at the end of this chapter.

The software development V-model presented in Chapter 2 is used as a basis for

the framework for its clarity and simplicity in presenting related software
development activities, not because it would describe how software is actually
developed. The activities shown in this model are generally acknowledged in
software development, but their interrelations and performance order may vary.

The requirements implementation framework is presented first at a detailed level

from the viewpoints of requirements analysis, designing, coding, and testing.
Requirements implementation elements in these activities are classified to
precondition, process, change management, and verification and validation
sections. The precondition section includes elements that need to be in order
before successful requirements implementation in the activity is possible. The

45
process section then presents elements that are utilised during the activity. While
the change management section discusses the important elements related to
requirements change, the verification and validation section concentrates on
elements that ensure the correctness of the produced work product in the
activity. After the individual activities with related requirements implementation
elements have been discussed, they are integrated into one holistic view.

The notation used in depicting the framework is shown in Figure 5.

Development activities, such as software requirements analysis, are shown as
rectangles. Work products from these activities are depicted as rounded
rectangles. Human figures describe roles, such as a software designer or a CCB.
Relations among activities, work products, and roles are shown as lines. A more
specific description of a relation is given inside or at the end of the line. The
relation can be either strong (solid line) or weak (dotted line). A strong relation
implies that the relation is essential from the viewpoint of requirements
implementation. A weak line indicates that the relation is sometimes
questionable and, for example, dependent on the nature of the development
project or product. Finally, a semi-dotted line compiles related elements
together.

Development activity Strong relation

Work product Weak relation

Role Compilation

Figure 5. Framework notation.

5.1.1 Requirements implementation in software requirements

analysis

Requirements implementation in software requirements analysis can be regarded

as an activity which transforms high-level requirements into low-level software
requirements and defines these requirements so that their further implementation
in the following development activities is possible. Requirements implementa-
tion elements in software requirements analysis are shown in Figure 6. These

46
precondition, process, and verification and validation elements according to
Section 3.1 and Chapter 4 are discussed next. Furthermore, change management
elements according to Section 4.1 are covered below.

- Documentation policy
- Traceability policy
- Change management
policy

- SW Developers

System
Requirements

- Customers

- Traceability Software - Define attributes

Requirements - Change procedure
- Consistency
Analysis - Baselining
- CCB
- SCM

Software - Review
Requirements
Specification - HW & SE
Developers
- Quality attribute
verification

Figure 6. Requirements implementation in software requirements analysis.

Precondition
To increase the success rate of a development project, system level requirements
development and allocation should include people with software expertise.
These people may be software analysts, designers, and testers. Software
developers' knowledge of the software's capability and cost-effectiveness to
implement a certain requirements may affect the decision to allocate the
requirement into either software or hardware requirement.

Before the implementation of system requirements into software requirements

can begin, certain working policies must be defined. The most crucial are a

47
requirements documentation policy, a change management policy, and a
traceability policy. These requirements policies need to be agreed to make sure
that requirements are later handled uniformly and correctly. For example, a
requirements documentation policy may define how requirements are classified
and identified in requirements documents. A traceability policy is needed to
agree and inform developers about which work products need to be traced and
how tracing is performed. A balance between the cost and the benefit of having
traceability information among work products should be determined. A change
management policy defines responsible roles and change procedures in
development activities when a change is introduced. Leaving requirements
policies undefined inevitably creates later a situation where working methods are
forgotten and not performed properly.

The above-mentioned requirements policies may be organisational-wide or

specific to a certain project. They may be even stated before system
requirements elicitation begins. Nevertheless, the definition of requirements
policies must take place before software development activities begin. It is also
important to inform the relevant developers about these policies. Developers
must understand the benefits and rationale behind certain working methods.
Only in this way will the developers be motivated to follow the defined
requirements policies.

Process
During the software requirements analysis, certain attributes for requirements
must be defined to enhance requirements implementation later. Requirements
need to have information about their identity, priority and stability. Identification
is needed for various elements of requirements implementation, such as
traceability, consistency check, and change management. Priority information is
needed to make design decisions. Valuable sources for requirements priority
information include customers and software developers. Requirements stability
information helps to control change of requirements. Defining a requirement's
stability may require software design and development expertise. In addition to
these attributes, documentation of requirements' source, scope and rationale may
be usable later.

48
Change Management
An unmanaged change in the requirements during software development may
easily cause the system to be inconsistent. The requirements analysis activity
must have a procedure for managing changes from system level and from lower
level development activities. This procedure has to make sure that the software
requirements are consistent with the system requirements and internal
consistency among the software requirements is preserved. To support impact
analysis of a change, traceability amongst the software requirements has to be
maintained. Software developers also have to have procedures for reporting and
negotiating with system engineering level about problems implementing certain
requirements in lower-level software development activities.

Baselining, or freezing a set of software requirements, is needed because

changing requirements cause too much instability in software development.
Although elicited requirements are not necessarily the final ones, a baseline of
requirements should be defined to enable consistent development of both
hardware and software. To baseline requirements and manage requirements
change, a change control board is needed. A CCB for embedded system
development may include representatives from software, hardware, and system
disciplines, and even people from hardware quality assurance and hardware
configuration management. For smaller and experimental projects, the CCB may
be very flexible and only consist of a few people. Both change management and
baselining are related to software configuration management, which controls the
change throughout the software development.

Verification and validation

The internal and external validity of the SRS must be ensured in a review.
Customers, software designers, and testers may take part in such review. Their
views on the adequacy of certain requirements may prevent approval of incorrect
requirements. Because the software is part of a larger embedded system, an
important purpose of the SRS is to determine the interfaces from software to the
system and hardware elements. Therefore, the presence of persons responsible
for system and hardware elements may be also required in the review. Finally,
the review of the SRS must include verification of quality attributes, which will
ensure that non-functional requirements have been sufficiently acknowledged in
the software requirements.

49
 5.1.2 Requirements implementation in software designing

Software design activity elaborates the software requirements into designs,

which can be used as blueprints to implement the desired software.
Requirements implementation in software designing includes elements that
allow developing correct and consistent designs against the requirements, and to
analyse and implement requirements changes. Both software architectural design
and detailed design activities include similar requirements implementation
elements; therefore, they are both discussed in this section. Requirements
implementation elements in software design are shown in Figure 7. The
precondition, process, and verification and validation elements covered in
Sections 3.2, 3.3, and Chapter 4 are discussed next. Also change management
elements covered in Section 4.1 are discussed.

Software - Preliminary designs

Requirements - Prototypes
Specification

- Trade-off analysis
- Traceability Software - Unrealized requirement
- Consistency Design procedure
- Change procedure - CCB
- SCM

- Review
Design - Coverage check

- Quality attribute
verification

Figure 7. Requirements implementation in software architectural and detailed

design.

50
Precondition
A good practice would be to include software designers in the specification of
software requirements, or even during the development and allocation of system
requirements. Their ability to evaluate and test proposed requirements in the
early phases of system and software development may be invaluable.
Demonstrating the system to the customers with preliminary designs and
prototypes is a cost-effective way of eliciting new essential requirements and
refining already agreed requirements. The results of preliminary designs and
prototypes could be also used later in the actual design activity as a basis for
design trade-off analysis.

Process
Design trade-off analysis ensures that requirements are implemented in the most
effective way and resolves conflicts when requirements collide. To make
decisions about which requirement is important to implement in the design, the
designer needs prioritisation information of requirements. Other information that
the designers may need from requirements include the rationale and source
behind a requirement. These requirement attributes enable designers to justify
design decisions, help to negotiate about fuzzy requirements, and make
requirements reuse possible. Specifying software requirements precisely and
quantitatively where possible during the requirements analysis activity pays off
now when designers have to make design decisions. However, requirements
should not be written as restricting design decisions, because they may prevent
designers discovering the most effective solution to a requirement.

Change management
Designers should have change procedures for making modifications in their
designs. These procedures may be jointly co-ordinated with the CCB and SCM.
Change procedures have to include impact analysis of the change for software
designs. Performing impact analysis produces much needed information for
project scheduling and resourcing. Without impact analysis a change of
requirements might result in a sequence of unsystematic changes to designs and
overrun of the project schedule.

Besides the defined requirements change procedure, software developers need a

procedure for unrealisable requirements. The developers will almost certainly

51
face ambiguity and conflicts in the requirements. Therefore, they need to report
about the problematic requirement and ask for clarification. An unrealised
requirement should finally be dealt by the CCB, which decides how the
requirement is changed.

Verification and validation

When software designs are ready for implementation, they need to be reviewed
for consistency with requirements. The review must also validate that all
requirements are covered. Requirements need to be individually identified to
enable consistency and coverage check. The same is expected in all other work
product reviews during software development. For some quality attributes, a
specific verification process is needed. This may be achieved, for example, by
defining and using checklists for specific quality attributes in design reviews.

5.1.3 Requirements implementation in software coding

In the software coding activity, software requirements are finally implemented in

a programming language that is understandable to the machine. Requirements
implementation in this activity is considered an activity to ensure that
requirements are properly realised. Preparing and analysing changes to software
code when software requirements change is also an important element.
Requirements implementation elements in software coding activity are shown in
Figure 8. These requirements implementation process and verification and
validation elements according to Section 3.4 and Chapter 4 are discussed next.
Furthermore, change management elements according to Section 4.1 in software
coding are covered below.

Process
The relationship between code and requirements is not as strong as is the case
with higher level designs. This is true especially when detailed designs are very
specific and enable automated code generation. On the other hand, if detailed
designs were not properly done, software requirements may support the
construction of code.

52
 Change management
Developers that are responsible for software implementation must be ready for
changes in the requirements. Procedures for change management need to be
defined before the development begins. Change procedures should be planned
and executed jointly with the CCB and SCM. Impact analysis is a significant
part of the change procedure. Because of the close relation between
implementation and detailed design, the impact analysis must consider both of
them simultaneously. For example, a change in detailed design might require
many changes in code, which in turn requires changes into detailed designs, etc.
Developers also need procedures for unrealised requirements similarly to those
in design activity. Some requirements' unfeasibility for implementation is not
revealed until the actual code is written and tested.

Software
Requirements
Specification

Software
Detailed - Change impact
Design analysis

- Change procedure
- Traceability Software - Unrealized requirement
- Consistency Coding procedure
- CCB
- SCM

- Review
Code

- Quality attribute
verification

Figure 8. Requirements implementation in software coding.

53
Although software standards (e.g. ISO/IEC 12207 and ISO/IEC 15504) require
traceability between software code and requirements, traceability links between
software code and software requirements is not always a necessity. If the code is
compatible with the detailed design, then all change management, consistency
checking, and other procedures concerning the requirements are dealt between
them. For example, the code may be automatically generated from detailed
designs, and all the needed changes are made to these designs. Once the changes
to the designs have been implemented, the code is regenerated. In this case,
traceability between requirements and the code is replaced by traceability
between detailed design and the code.

Verification and validation

Finally, when a functionality, a component, or some other part of the
implementation is ready, it needs to be reviewed for correctness. If the software
code is transformed from the detailed designs, its consistency with the designs
must be ensured. Otherwise the code's consistency directly with the software
requirements must be checked. A software code review process may also include
a verification process for specific quality attributes.

5.1.4 Requirements implementation in software testing

Requirements implementation in testing refers to the elements that support

testing the software rigorously against the requirements. Software testing
consists of sequential activities such as module, integration, and system testing.
Requirements implementation elements are mostly related to system testing.
These elements and their relation to software testing are shown in Figure 9. The
precondition, process, and verification and validation elements of requirements
implementation covered in Section 3.5 and Chapter 4 are discussed next. Also
change management elements covered in Section 4.1 are covered below.

Precondition
Requirements must be stated unambiguously and quantitatively to enable
development of correct test cases. The SRS must have validation criteria for
accepting certain requirements. These criteria should be defined as precisely as
possible. Testers may already start planning test cases while software requirements

54
 are being elaborated. Writing test cases for requirements during their development
reveals possible flaws inexpensively.

- CCB, SCM,
TCM
- Test case
development

- Change procedure
Software - Coverage check
- Quality attribute
Requirements - SRS verification
Software verification
Specification - Review
System
Test Testing
Cases

- Traceability - System &

- Consistency HW testers
- Validation criteria Software
Integration
Testing

Software
Module
Testing

Figure 9. Requirements implementation in software testing.

Process
Module and integration tests ensure that requirements stated for architectural and
detailed designs are correctly fulfilled. Module testing is often performed
simultaneously with the coding activity. For some systems it may be necessary
to trace these lower level tests to the requirements and check that all behaviour
and quality attributes are covered. However, if software designs' traceability and

55
consistency with requirements are maintained, then there should not be any need
for direct linkage between requirements and module and integration test cases.

The purpose of system testing is to ensure that the developed software works
correctly in its actual environment and reflects correctly system and hardware
requirements. Therefore, embedded software testers may need to have co-
operation with system and hardware testers throughout software testing.

Change management
Since software testing is strongly related to requirements, a change of a
requirement also affects testing activities. To maintain consistency throughout
the development chain, testing activity needs procedures in case of a
requirement change. For example, a change procedure may include an
evaluation of what test cases in system testing need to be updated. Maintaining
traceability between test cases and requirements facilitates this kind of
evaluation. If module and integration tests are also dependent on requirements,
then change procedures need to be defined for them as well. The CCB, SCM,
and test case management (TCM) play a vital role in defining how the change is
implemented.

Verification and validation

Checking consistency between requirements and test cases, for example in a
review, will ensure that correct aspects are being tested. Testing of some quality
attributes may require specific testing arrangements. For example, testing of
usability may require usability tests that include users. Maintaining test cases
traceability with requirements enables to check that all requirements are covered.
Furthermore, satisfactory implementation of requirements can be proven to
customers, when system test results are traceable to customer requirements.

5.1.5 Requirements implementation throughout the development

The previous sections discussed the elements of effective requirements

implementation and interrelations of these elements in separate software
development activities. In this section, a summary is made by combining the
most essential elements into one holistic view. These elements should compose a

56
chain of supporting activities for software development that enable correct and
effective implementation of requirements. Implementation of requirements
throughout the development is shown in Figure 10.

Software developers should be considered before the actual software

development begins. Their involvement during system requirements
development may save the project from expensive changes in later development
phases. Also documentation and working policies regarding requirements in
software development need to be defined. Having clearly defined policies and
informed and motivated developers on these policies will ensure successful
handling of requirements.

One of the most important elements of successful requirements implementation

is obviously tracing software requirements backwards to system requirements
and forwards to software work products. Traceability is needed for requirements
change management, consistency assurance, and coverage checks throughout the
development. In the proposed framework, traceability covers all development
activities except for coding, module, and integration testing. To introduce
traceability into the software development successfully, it should elaborate
seamlessly into software process through methods, techniques, and notations
used in requirements analysis, design, code and test. Indeed, this requires a clear
traceability policy, requirements identification and automation of traceability
information maintenance.

57
 System - Documentation policy
Requirements - Traceability policy
Analysis - Software - Change management policy
Developers - Review practices

Software Software Software

- Define Requirements - Validation
Requirements criteria System
attributes Specification
Analysis Testing

- Traceability
- Consistency
- Coverage - Change
check Software procedure Software
Architecture - CCB Integration
Design - SCM Testing

Software Software
- Trade-off Detailed Module
analysis Design Testing

- Change impact
analysis
- Unrealized Software
requirement Coding
procedure

Figure 10. Requirements implementation throughout the development.

Requirements traceability is more challenging to implement in an iterative-like

software development process than it is in a waterfall-like process. The waterfall
model's sequential and clear order of activities helps to add traceability
information to different baselines. However, the benefits of traceability are
significant in iterative development where the change of designs,
implementations, and test cases is continuous. Regardless of the development
process model, ensuring requirements consistency and coverage during

58
development activities is facilitated, if the development organisation traces
requirements to work products and has established review practices in use.

Another important element of requirements implementation is management of

requirements change throughout the development. To avoid uncontrolled
changes in the developed system and software, the organisation should have a
CCB that manages changes. Because of the close relationship between
requirements change management and SCM, their integration may bring
valuable benefits. Requirements change management should include change
management policy and change procedures for separate activities. Stability of
requirements must be known to prepare designers for requirements change. In
addition, impact analysis, in case of a change, needs to be done in design and
coding activities. During these activities developers must also perform trade-off
analysis and be ready for unrealisable requirements.

The proposed framework assumes that software code can be directly created
based on detailed designs. Thus, traceability and consistency between the code
and the requirements is maintained through detailed designs. The framework
also assumes that module and integration tests are based on detailed and
architectural designs, and that their mutual traceability and consistency is
managed without any direct relationship with requirements. These and other
assumptions in the framework may be different from an organisation's actual
software development process. Therefore, possibilities to tailor the framework
for different development projects and organisations are discussed in the next
section.

5.2 Adaptation of the framework

Constructing a requirements implementation framework that is suitable for

analysing and improving requirements implementation in different kinds of
development environments, life cycles, and processes is not an easy task.
Furthermore, to develop a framework that covers all the essential
implementation elements, but is not at the same time overly complex, presented
problems. The relationships and dependencies among software development
activities and requirements implementation elements are clearly not
straightforward.

59
The need and ability to sacrifice resources to requirements implementation is
dependent on both the size of the developing company and the size of the
project. More strict approach is suitable for large companies and large complex
projects that have a number of teams in different disciplines developing the same
product. In contrast, small companies and small projects need a more informal
and flexible approach. Small and large companies' needs and resources are
obviously not comparable. In a small organisation, for example, reporting an
unfeasible requirement may be informal communication between a developer
and a project manager. Nevertheless, roles and procedures regarding
requirements implementation in both small and large organisations should be
clearly defined.

The nature of developed embedded product also governs the attitude towards
requirements implementation aspects. Product and also project requirements
direct how the software development is carried out. For example, development
of a safety critical system may require an extremely rigorous approach in the
implementation of the given requirements. In contrast, an experimental project
may have very flexible procedures for requirements implementation. The
demands for the software process in the development of an aircraft are quite
different from those in the development of an electronic alarm clock.

To enhance requirements implementation, the proposed framework's elements

need to be adapted for the project's development life cycle. The elements and
their interrelations can be tailored to serve as a good software practice for
incremental, evolutionary, or any other development life cycle. For example, if
the developed product requires a risk driven development approach such as
spiral development model (Boehm 1988), the requirements implementation
elements shown in the framework may be iterated with the development
activities. The purpose of the framework is not to supplement the traditional
sequential development life cycle's deficiencies on requirements implementation
concepts. On the contrary, it should be tailored for the needs of any life cycle,
project, and product.

Before requirements implementation in software development can be improved,

the system requirements engineering processes must be in order. Stakeholders
needs must be understood and conformed. This requires a firm grasp of the
requirements elicitation and validation processes. Also traceability and other

60
requirements management elements must be in order when stakeholder needs are
developed into system requirements and further allocated into software
requirements.

Literature study revealed that developers' motivation to use some of the

framework's elements during software development is hard to achieve. For
example, elaborating effective traceability and consistency check practices into a
software development process is regarded as time-consuming and expensive.
Gaining developers' acceptance of these aspects would be very challenging.
They are already over employed and certainly do not need more bureaucracy
that does not explicitly help doing their work. By explaining the rationale behind
requirements implementation elements for developers, the chance for their
commitment to these is increased.

61
 6. Validation of the requirements
implementation framework
Two case studies were conducted to validate the requirements implementation
framework's capability to analyse and improve requirements implementation
practices in an industrial environment. This chapter presents these studies and
discusses the validation of the framework. First, the research approach used in
the validation is overviewed. Then, the two case studies are discussed in more
detail and the most important findings are presented. Finally, the framework's
applicability is evaluated based on the results from the case studies.

6.1 Research approach for validation

Software engineering as a science discipline should involve an experimental

component to test or disprove suggested theories. Researchers cannot rely
merely on conclusions following their logical thought. Indeed, experiments have
to be made with models, methods, and techniques to find out how and when they
really work, to realise their limits, and to improve them. (Basili 1996.)

Experimentation of the suggested model in this study was achieved with an

historical research approach according to Zelkowitz and Wallace's (1998)
categorisation of software engineering validation methods. Historical research
approach allows researchers to study organisations' legacy data and lessons
learned by exploring existing software artefacts and by interviewing personnel.
This approach was used in this study to evaluate the current state of
requirements implementation practices in the case organisations.

The sheer historical validation approach was further extended by suggesting

improvements to case organisations' requirements implementation practices
based on the framework and by having the organisations' personnel to evaluate
these suggestions. The framework was found to be useful if it could be
successfully used in discovering problems in case organisations' current
requirements implementation practices and suggesting valuable improvements to
these problems. This condition was evaluated by analysing case organisations'
feedback on suggested improvement proposals.

62
Data for the current state analysis of requirements implementation practices in
the two case organisations was collected from multiple sources. First, problem
areas of requirements implementation in the organisations were surveyed with
an open ended questionnaire based on the questions in Appendix 1. Then, the
discovered problem areas were further examined with more detailed
questionnaires, interviews, and by investigating organisations' software related
documentation.

By comparing the framework's elements and problem areas in a case

organisation's current practices, an improvement proposal was composed. If a
case organisation's development process lacked an element from the framework,
its negative consequences were examined in more detail and a possible solution
was considered. All solutions were then combined into an improvement
proposal. This included processes, methods, and tools that the researcher
believed would enhance requirements implementation practices. The
summarised list of requirements implementation practices and methods in
Appendix 2 was used as the basis for concrete improvement suggestions.

The improvement proposal was then reviewed by case organisations to get

feedback on improvement suggestions. This feedback was used as evidence for
validation of the requirements implementation framework's applicability. The
validation of the framework was augmented by having two different kinds of
embedded software developing organisations as subjects. The first studied case
organisation is a large company that develops various products and has hundreds
of people developing software. The second organisation is considerably smaller
with several dozen developers and is focused on developing few products.

6.2 Case one - large organisation

The first case organisation builds embedded systems for document processing.
The organisation's R&D consists of over 2000 people, including some 700
software developers. Software development approaches include application,
control, and embedded software. The software is developed with a tailored
development model from the classical V-model. Currently, the software
development is under a process improvement program and the goal is to
introduce CMM level 2 key process areas.

63
Requirements are supplied by marketing, maintenance, and R&D departments.
These requirements are elaborated with safety rules, legislation, and standards
into more detailed product specification. At the product level, requirements are
initially specified in natural language and then developed into use cases, and
sequence diagrams. Product requirements are decomposed into sub-system
requirements, which contain requirements for mechanical, electronic, and
software disciplines. This decomposition can continue to more detailed sub-
systems, until requirements are implemented in mono-disciplinary levels.

For this study, requirements implementation practices in a software development

unit of the R&D were explored. The unit's software integrator acted as a source
of information for the inquiries. Also documents such as software process
descriptions, SRS templates, and inspection checklists were examined.

6.2.1 Current state analysis

Software level requirements documentation in the case organisation was well

managed and performed. For example, the organisation had defined a
requirements management policy and a template for requirements specifications.
The SRS template, however, lacked individual requirements' attributes such as
stability, rationale, source, and priority. But these insufficiencies had already
been noticed and improvement actions had been initiated. Requirements
management was also planned to be improved with a commercial tool.

Internal verification of SRS's and baselining the requirements were also well
handled. Consistency checking between software requirements and other
software work products was also conducted. All development activities included
review practices, where consistency was ensured. Checklists and other
guidelines for the reviews imposed to check internal and external consistency of
the work products.

Another well-controlled aspect, quite surprisingly, was requirements change

management. Although the organisation reported having changing requirements as
quite common phenomenon, it had established such change management practices
that these changes did not have a negative effect on development activities or

64
work products. The organisation had established a change management process
with a change management tool and a change control board.

Clearly the most serious problems that hinder requirements implementation in

the organisation were with system requirements development and management.
Although software requirements were well documented and managed, system
requirements tended to be documented insufficiently and managed poorly. There
was no system level requirements management process nor was there a common
requirements documentation policy. This made tracing of software requirements
to system requirements difficult. Furthermore, insufficient participation of
software engineers in system requirements development and allocation lead to
problems later when software development begins.

Problems also seemed to arise because of inadequate co-operation among

software, hardware, and system levels. For example, informal checking of
software requirements consistency with system requirements had later caused
problems in system integration and testing because software and system
requirements were not consistent after all.

Although software requirements change management was well managed, impact

analysis of a change sometimes caused problems. Impact of a change was
estimated by team-leaders before approval in the CCB. Sometimes these
estimates were too vague and caused problems for project schedules. Other
minor problems related to requirements implementation included the software
architecture's analysis against requirements, testing of quality attributes, and lack
of traceability information from requirements to design and code.

6.2.2 Improvement proposal

In order to improve requirements implementation in software level, the

organisation must pay attention to system requirements development, allocation,
and management. System requirements development and allocation procedures
must include software engineers who could use measurement information and
prototypes to specify development cost of certain software solution.
Documentation and management of system requirements must be also enhanced.
Organisation wide requirements policy containing procedures and templates

65
concerning all disciplines would enhance development, communication, and
reuse of requirements. A common policy would also allow the use of one
requirements management tool for maintaining all requirements.

Involvement of other disciplines in verification of software requirements should

be improved. Software requirements must be consistent and compatible with
hardware right from the beginning. In the past, hardware engineers were not
involved in SRS reviews at all. Fortunately, this situation is now changing and
other disciplines are more involved in these reviews. The current practice of
informal reviews may be evolved towards more formal inspections with reading
techniques such as PBR to make the involvement even more effective.

Implementation of requirements in the software's architecture could also be

improved. Software's architectural design is currently very much dependent on
the experience of the software architects. The architects manage to get the
system working, but whether it can be developed to be the most efficient
solution is another issue. Therefore, the architecture's trade-off and quality
analysis methods such as ATAM, SBAR, and AQA could be utilised.

A few improvements in the coding and testing activities could be also made.
Firstly, static code analysers could be used to identify areas of the source code
that are not conformant to certain quality attributes. Analysers such as Logiscope
and QAC/++ would automatically discover problems in maintainability,
readability, and portability issues. Secondly, the requirements specification's
quality attributes could be quantified and verified already in the software level
tests rather than postponing quality attribute testing to the system level.

Finally, traceability information from software requirements to designs and code

could be defined. Tool or matrix-based tracing would enhance the consistency
and coverage checking process between requirements, designs, and code.
Furthermore, tracing could improve current problems with an impact analysis of
a change.

66
 6.3 Case two - small organisation

The second case organisation develops data management solutions for embedded
devices. The organisation employs around 100 people. One-fourth of these
employees work in R&D. Development projects at the organisation are carried
out at one site with an incremental software development process. The
organisation has an ongoing software process improvement program that is
based on ISO/IEC 15504 standard.

Requirements come from different sources such as customers, sales and

marketing departments, standards, and product management. These requirements
are translated into an external specification written in natural language. The
external specification then serves as a contract between stakeholders and
developers. The original stakeholder requirements are not documented in a
separate requirements document. In case of a new feature, the external
specification is transformed into more detailed software requirements.

The organisation's director of R&D served as source of information for the case
study. In addition, documents such as software process descriptions, requirement
specification templates, and review guidelines were examined.

6.3.1 Current state analysis

The case organisation's requirements documentation was sufficiently well

managed. Documentation was seen as useful and guidelines for the
documentation were used. Also templates for requirements specifications were
available. The requirements were properly classified and included appropriate
attributes. Requirements documentation also included verification criteria for
necessary non-functional requirements.

Requirements implementation in the coding and testing activities was also well
handled. The coding activity included review practices for certain parts of the code
where requirements implementation needed to be evaluated. Competent
developers ensured absorption of quality attributes into the code, such as
portability. Maintainability of the code was achieved by following the
organisation's coding standards. The testing activity included a set of testing tools

67
that automatically ensured verification of correct realisation of requirements.
Sometimes, however, automatic testing was not possible. In such case, manual
work and tracking of test cases to requirements was required. Tracking between
requirements and test cases was being introduced more and more.

The organisation was not experiencing many problems with changing

requirements. A product manager managed the change process and changes were
handled at weekly meetings. The organisation had established requirements
baselining practice in use and had a change approval body similar to the change
control board. Also software configuration management practices were in use to
control change. The organisation planned to introduce a tailored requirements
management tool to enhance requirements management and implementation.

Prioritisation of requirements was seen as an important factor for the whole

development process, but questions arose about the effectiveness of the current
prioritisation process. Currently, requirements prioritisation with various
stakeholders is performed at meetings and is mainly based on intuition. Whether
all necessary requirements were covered and whether they were correctly graded
was not so evident.

Review practices in the organisation were not fully taken into use. When reviews
were performed, they tended to follow a more informal than formal process.
Limited resources simply do not allow extensive check of consistency between
requirements and other work products.

Other insufficiencies were observed with requirements tracing and architecture

designing. It was not always clear how to trace lower-level requirements to test
cases. Architecture designing was reported to be well performed, but it was
mainly based on designers' expertise and architecture analysis against quality
attributes was not commonly performed.

6.3.2 Improvement proposal

To improve current practice of requirements prioritisation, the organisation

should take a prioritisation method into use. Obviously, the method will not
solve all the prioritisation problems by itself, but it would give a more

68
rationalised basis for making priority decisions. One possible solution would be
Wiegers' prioritisation method that is based on evaluating requirements priority
from the viewpoint of different stakeholders and developers.

Another improvement area to consider would be the current practice of reviews.

More formal inspections may remove defects more efficiently and enhance
consistency between requirements and work products. Extensive reviews in all
development activities, however, are not a reality in small organisations.
Therefore, a root cause analysis must be done to locate where defects have
infiltrated and to improve review practices in the most vulnerable development
activities. Automation of certain reviews would minimise developers'
involvement. For example, code reviews may be supplemented with a static code
analysis tool such as QAC/++ that verifies portability of the code.

Other improvements may be accomplished by using the upcoming requirements

management tool to trace requirements to test cases and by utilising architectural
analysis methods more commonly. Linking test cases to requirements with the
requirements management tool would make it possible to prove when all
features have been covered in the testing activity. Although architecture
designing was well performed, architecture trade-off and quality analysis
methods could be used to enhance evaluation of architecture against quality
attributes.

6.4 Applicability of the requirements implementation

framework

After the current state analyses and the improvement proposals on requirements
implementation for the case organisations were made, the organisations
evaluated the significance of the given improvement proposal. The organisations
assessed the proposal's individual improvement suggestions significance with a
five-grade scale, where the lowest grade indicated an insignificant improvement
and the highest grade a critical improvement. This feedback information was
used to evaluate the requirements implementation framework's applicability in
improving requirements implementation practices.

69
The improvement proposal for the first case organisation was considered to be
very significant. A total of seven individual improvement suggestions were
made. Two of these suggestions were considered critical. Both of these
concerned system requirements development and documentation. Furthermore,
three suggestions were considered important. These suggestions included the
involvement of other disciplines in software requirements reviews, architectural
analysis against requirements, and static code analysis. The rest of the
suggestions - quality attribute quantification and post-traceability - were seen as
less important and did not therefore require immediate actions. The results from
the large organisation's feedback indicated that the framework was very useful in
revealing weak spots and developing improvement suggestions.

The second organisation received total of five improvement suggestions. The

organisation evaluated two suggestions to be important and the rest of the
suggestions as less important. The current state with requirements prioritisation
and work product inspections were considered to be important improvement
areas. Static code analysis and architecture evaluation were considered less
important improvements and their utilisation would require a closer study to see
the real benefits. Finally, tracing test cases to requirements with the requirements
management tool was not currently considered to be feasible.

The improvement proposal made for the large organisation was more successful
compared to the proposal made for the small organisation. One of the reasons for
this is the fact that the large organisation had much clearer system and hardware
levels in its development life cycle. Therefore, improvement areas in the
framework that concerned those levels were not that applicable in the small
organisation. Furthermore, the large organisation with a rigid and document
oriented development process had naturally more need for the elements
presented in the framework. Nevertheless, the improvement suggestions made
for the small organisation were also considered important and therefore the
framework can be also regarded as applicable for analysing and improving small
organisations' requirements implementation practices.

70
 7. Conclusions
This chapter concludes the study of requirements implementation in embedded
software development. First, the research is summarised by answering the
research questions. Then, the significance of the results is discussed. Finally,
further research possibilities based on this study are given.

7.1 Answers to the research questions

This research studied the means to improve requirements implementation in

embedded software development. The goal of the research was to gather
elements for effective requirements implementation from the literature and
integrate these into a common framework. The framework could be then used in
analysing and improving organisations' requirements implementation practices.
The goal was achieved by answering these research questions as set out in the
introductory chapter:

1. How can requirements be effectively implemented in embedded software

development?
1.1. What is the relationship between requirements and development
activities?
1.2. How should requirements be implemented during development?
1.3. What kind of framework would help to analyse and improve
requirements implementation practices?

The first sub-question was answered in Chapter 3 by discussing the impact of

various requirements on different development activities. It was shown that
requirements manifest themselves clearly in all development activities and that
quality attributes in particular are problematic to implement and verify.

Chapter 3 also partly provided answers to the second sub-question by presenting

requirements implementation methods for distinct development activities. The
second sub-question was further covered in Chapter 4, where requirements
implementation-supporting elements were discussed. These elements included

71
requirements change management, traceability, and consistency checking, which
are all visible throughout the development.

In Chapter 5, the requirements implementation elements found in Chapters 3 and

4 were gathered into a common framework in order to answer the third sub-
question. The developed requirements implementation framework was intended
to be used as a means of analysing and improving embedded software
organisations' capability to implement requirements effectively. This hypothesis
was validated in Chapter 6 which discussed the case studies in the small and the
large organisation.

The results from the case studies showed that the framework was indeed
applicable in analysing and improving requirements implementation practices in
an actual industrial environment. Most of the given improvement suggestions for
the large organisation were considered to be important or even critical. Part of
improvement suggestions made for the small organisation were also considered
important.

7.2 Significance of the results

The main result of this study is the requirements implementation framework for
embedded software. The framework is plainly a synthesis of previous research
results and used practices in the industry. What, then, is so novel and non-
obvious about their integration? The main contribution of the framework is that
it covers relevant elements as extensively as possible. It also represents
relationships among these elements to increase knowledge about their
interdependencies. Other meaningful results from this study are the
questionnaire, which can be used to clarify an organisation's current state of
requirements implementation practices, in Appendix 1, and references to the
concrete processes, methods, and tools for these practices in Appendix 2. These
both can be used as a basis for software process improvement from the
viewpoint of requirements implementation.

The developed framework's strength and at the same time its weakness is that it
covers a great deal of different software development areas and even system
engineering aspects. This caused the framework to evolve into a fairly general

72
model. One element of the framework, for example requirements change
management, could have formulated its own framework. On the other hand,
requirements implementation could have been examined solely from one
software development activity's viewpoint such as designing or coding. This
would have made possible more detailed analysis and improvement suggestions
on more specific areas.

However, general models are also needed. As was shown, the elements of the
framework affect each other from the system requirements level down to
software testing level. Therefore, the developed framework could identify
improvement areas from a large perspective. Indeed, an overall picture of the
problem domain is needed to isolate specific improvement areas. What then
separates the framework from known software process improvement reference
models such as CMMI or ISO 15504? This study's framework should be used
especially when requirements implementation causes problems in development
activities. Therefore, it may be used to complement software process
improvement efforts that are based on the reference models, as was shown in the
case studies.

After the framework was elaborated, its applicability in actual software

development environment was validated. The main problem with the validation
was the low number of case studies. Is it legitimate to draw conclusions about
the framework's applicability from only two case studies? Furthermore, a case
study may contain so many organisation-specific factors that it may be
impossible to derive generic results (Potts 1993).

The results from the two case studies indicated that the framework was indeed
useful. Two different kinds of organisations - a large one and a small - were
studied to augment the validity of the results. In addition to the missing practices
found in the organisations, the framework also revealed many well performed
elements. This also supports the claim that the framework includes essential
elements for effective requirements implementation. However, more case studies
would be needed to substantiate the results. Conducting such studies is by no
means an easy task, because a proper case study requires a very close
examination of the subject or phenomenon. In this study, it meant examining a
significant number of software documents, preparing and analysing
questionnaires, and interviewing software developers.

73
The developed framework provides the means to obtain an overall picture of
embedded software development and locate possible improvement areas related
to requirements implementation. Although the framework was intended for
embedded software, it is also applicable to analysing non-embedded software
development processes. In this case, system engineering and hardware related
elements from the framework may be ignored.

7.3 Further research possibilities

The requirements implementation framework's usefulness was validated in

embedded software environment by making a current state analysis and
proposing improvement suggestions. However, implementation of the proposed
suggestions and their possible benefits were not studied in this research. Such a
study should be conducted to evaluate the actual value of the framework based
improvement suggestions.

The developed framework was shown to be an effective foundation for improving

a large organisation's requirements implementation practices. More research is
needed to tailor the framework so that it is more capable of improving a small
organisation's problems cost-effectively. Further studies are also needed to clarify
what kinds of requirements implementation elements are essential to the
development of non-embedded software such as information systems.

The results from the case studies implied that while software requirements were
reasonably well handled, there were serious problems with system requirements.
In order to improve this, the participation of software developers at the system
engineering level should be studied more carefully. For example, software
measurement information and prototypes may be used to evaluate development
costs and rationalise system requirements allocation procedures.

74
 References
Arnold, R.S. & Bohner, S.A. 1993. Impact Analysis - Towards a Framework for
Comparison. Proceedings of the Conference on Software Maintenance 1993, 292 –301.

Basili, V.W. 1996. The Role of Experimentation in Software Engineering: Past,

Current, and Future. Proceedings of the 18th International Conference on
Software Engineering, 1996, 442 –449.

Boehm, B. W. 1988. A Spiral Model of Software Development and

Enhancement. IEEE Computer, Vol. 21, no. 5, 1988, 61 –72.

Chechik, M. & Gannon, J. 1995. Automatic Analysis of Consistency between

Implementations and Requirements: a Case Study. Proceedings of the Tenth
Annual Conference on Computer Assurance, 1995, 123 –131.

Chechik, M. & Gannon, J. 2001. Automatic Analysis of Consistency between

Requirements and Design. IEEE Transactions on Software Engineering. Vol. 27,
no. 7, 2001, 651–672.

CMMI-SW. 2002. Capability Maturity Model® Integration for Software

Engineering, Continuous Representation, Version 1.1. Available:
http://www.sei.cmu.edu/pub/documents/02.reports/pdf/02tr001.pdf
[Referenced 2.6.2003].

Crnkovic, I., Funk, P. & Larsson, M. 1999. Processing Requirements by

Software Configuration Management. Proceedings of 25th EUROMICRO
Conference, 1999. Vol. 2, 260–265.

Davis, A.M. 1988. A Comparison of Techniques for the Specification of

External System Behaviour. Communications of the ACM, Vol. 31, no. 9, 1988,
1098–1115.

Davis, A.M. 1990. Software Requirements: Analysis and Specification. New

Jersey, USA: Prentice-Hall.

75
DOORS 2003. Telelogic products - DOORS. Available:
http://www.telelogic.com/products/doorsers/doors [Referenced 15.11.2003]

Dorfman, M. 1997. Requirements Engineering. In Thayer, R.H. & Dorfman, M.

(eds.) 1997. Software Requirements Engineering, 2nd ed. Los Alamitos, USA:
IEEE Computer Society Press.

Douglass, B.P. 2000. Real-Time UML, 2nd ed. Reading, USA: Addison-Wesley.

Easterbrook, S. 2001. Lecture slides on Software Lifecycles. Available:

www.cs.toronto.edu/~sme/CSC444F/slides/L04-Lifecycles.pdf [Referenced 20.5.2003].

Ebert, C. 1997. Dealing with nonfunctional requirements in large software systems.

Annals of Software Engineering 3, Kluwer Academic Publishers, 367–395.

Ernst, R. 1998. Codesign of Embedded Systems: Status and Trend. IEEE Design
& Test of Computers, Vol. 15, no. 2, 45–54.

Gajski, D.D. & Vahid, F. 1995. Specification and Design of Embedded

Hardware-Software Systems. IEEE Design & Test of Computers, Vol. 12, no. 1,
53–67

Gotel, O.C.Z. & Finkelstein, C.W. 1994. An Analysis of the Requirements

Traceability Problem. Proceedings of the First International Conference on
Requirements Engineering, 94–101

Gross, D. & Yu, E. 2001. From Non-Functional Requirements to Design

Through Patterns. Requirements Engineering, Vol. 6, no. 1, 18–36.

Grünbacher, P., Egyed, A. & Medvidovic, N. 2001. Reconciling Software

Requirements and Architectures: The CBSP Approach. Proceedings Fifth IEEE
International Symposium on Requirements Engineering, 202–221.

Gupta, P. & Cunning, S.J. & Rozenblit, J.W. 2001. Synthesis of High-Level
Requirements Models for Automatic Test Generation. Proceedings of the Eighth
Annual IEEE International Conference and Workshop on the Engineering of
Computer Based Systems, 76–82.

76
Harker, S.D.P., Eason, K.D. & Dobson, J.E. 1993. The Change and Evolution of
Requirements as a Challenge to the Practice of Software Engineering.
Proceedings of the IEEE International Symposium on Requirements Engineering
1993, 266–272.

Hofmeister C., Nord, R. & Soni, D. 2000. Applied Software Architecture.

Reading, USA: Addison-Wesley.

IEEE Std 830–1998. Recommended Practice for Software Requirements

Specification. 1998. Institute of Electrical and Electronics Engineering, Inc.

Inverardi, P., Muccini, H. & Pelliccione, P. 2001. Checking consistency between

architectural models using SPIN. Proceedings of International Conference on
Software Engineering 2001 Workshop “From Software Requirements to
Architectures” (STRAW'01), 62–67.

ISO/IEC 12207. 1995. Information technology - Software lifecycle processes.

ISO/IEC.

ISO/IEC 15504. 1998. Information technology - Software process assesment -

Part 5: An assesment model and indicator guidance. Draft technical report.
ISO/IEC.

Järvinen, P. & Järvinen, A. 2000. Tutkimustyön metodeista. Tampere, Finland:

Opinpaja.

Kandt, R.K. 2002. Software Configuration Management Principles and Best

Practices. In: Oivo, M. & Komi-Sirviö, S. (eds.) 2002, Product Focused
Software Process Improvement. 4th International Conference PROFES 2002,
LNCS 2559, Springer-Verlag, 300–313.

Karsai, G., Sztipanovits, J., Ledeczi, A. & Bapty, T. 2003. Model-Integrated

Development of Embedded Software, Proceedings of the IEEE, Vol. 91, no. 1,
January, 145–164.

Knight, J.C. & Myers, E.A. 1993. An Improved Inspection Technique.

Communications of the ACM, November 1993, Vol. 36, No. 11, 51–61.

77
Kotonya, G. & Sommerville, I. 1998. Requirements Engineering, Processes and
Techniques. Chichester, England: John Wiley & Sons Ltd.

Kozlenkov, A. & Zisman, A. 2002. Are their Design Specifications Consistent

with our Requirements? Proceedings of the IEEE Joint International Conference
on Requirements Engineering 2002, 145 –154.

Lam, W. & Shankararaman, V. 1999. Requirements Change: a Dissection of

Management Issues. Proceedings of 25th EUROMICRO Conference, 1999. Vol.
2, 244–251.

Lee, E.A. 2000. What's Ahead for Embedded Software? IEEE Computer, Vol.
33, no. 9, September 2000, 18–26.

Lindström, D.R. 1993. Five Ways to Destroy A Development Project. IEEE

Software, Vol. 10, no. 5, September 1993, 55–58.

Liu, L. & Yu, E. 2001. From Requirements to Architectural Design - Using

Goals and Scenarios. ICSE-2001 Workshop: From Software Requirements to
Architectures. Available: http://www.cs.toronto.edu/pub/eric/STRAW01-
R2A.pdf [Referenced 20.5.2003].

Logiscope. 2003. Telelogic Tau Logiscope. Available:

http://www.telelogic.com/products/tau/logiscope/ [Referenced 17.9.2003].

López, J.C., Hermida, R. & Geisselhardt, W. 1998. Advanced Techniques for

Embedded Systems Design and Test. Dordrecht, The Netherlands: Kluwer
Academic Publishers.

Mazza, C., Fairclough, J., Melton, B., De Pablo, D., Scheffer, A., Stevens, R.,
Jones, M. & Alvisi, G. 1996. Software Engineering Guides. Prentice Hall.

Mead, N.R. 1994. The Role of Software Architecture in Requirements Engineering.

Proceedings of the First International Conference on Requirements Engineering
1994, 242.

78
MOOSE 2002a. Survey of Applicaple Architectural Design and Analysis
Methods. MOOSE Consortium.

MOOSE 2002b. Industrial Inventory Report. MOOSE Consortium.

Mäkäräinen, M. 2000. Software Change Management Processes in the

Development of Embedded Software. VTT Publications 416. Espoo, Finland:
Technical Research Centre of Finland.

O'Neal, J.S. & Carver, D.L. 2001. Analysing the Impact of Changing
Requirements. Proceedings of IEEE International Conference on Software
Maintenance, 2001, 190–195.

Olsson, T. & Grundy, J.C. 2002. Supporting Traceability and Inconsistency

Management Between Software Artifacts. Proceedings of the IASTED
International Conference on Software Engineering and Applications 2002,
Boston, USA.

Olsson, T. & Runeson, P. 2002. Document Use in Software Development: A

Qualitative Survey. Software Engineering, research and practise in Sweden 2002.
Available: http://serg.telecom.lth.se/research/publications/docs/100_olsson_
runeson_document_use.pdf. [Referenced: 5.6.2003].

Palmer, J.D. 1997. Traceability. In Thayer, R.H. & Dorfman M. (eds.) 1997,
Software Requirements Engineering, Vol. 2. Los Alamitos, USA: IEEE
Computer Society Press. 364–374.

Parviainen, P., Hulkko, H., Kääriäinen, J., Takalo, J. & Tihinen, M. 2003.
Requirements Engineering, Inventory of technologies. VTT Publications 508.
Espoo, Finland: Technical Research Centre of Finland.

Potts, C. 1993. Software-Engineering Research Revisited. IEEE Software. Vol.

10, no. 5, September 1993, 19–28.

Pressman, R.S. 2000. Software Engineering, A Practitioner's Approach, European

Adaptation, 5th ed. Berkshire, England: McGraw-Hill.

79
Punter, T., Trendowicz, A. & Kaiser, P. 2002. Evaluating Evolutionary Systems.
In Oivo, M. & Komi-Sirviö, S. (eds.) 2002, Product Focused Software Process
Improvement. 4th International Conference PROFES 2002, LNCS 2559,
Springer-Verlag, 258–271.

QAC/++. 2003. QAC/++. Available:

http://www.programmingresearch.com/solutions/qac3.htm [Referenced
17.9.2003].

Ramesh, B., Powers, T., Stubbs, C. & Edwards, M. 1995. Implementing

Requirements Traceability: a Case Study. Proceedings of the Second IEEE
International Symposium on Requirements Engineering 1995, York, England, 89–95.

Rational RequisitePro 2003. Rational RequisitePro - Product overview.

Available: http://www-306.ibm.com/software/awdtools/reqpro [Referenced:
15.11.2003].

Robinson, W.N. 2002. Monitoring Software Requirements using Instrumented

Code. Proceedings of the 35th Hawaii International Conference on System
Sciences 2002, 3600–3609.

Robinson, W.N. & Pawlowski, S.D. 1999. Managing Requirements Inconsistency

with Development Goal Monitors. IEEE Transactions on Software Engineering,
Vol. 25, no. 6, 1999, 816–835.

Rodríguez-Dapena, P., Vardanega, T., Trienekens, J. & Brombacher, A. 2001.

Nonfunctional Requirements as a Driving Force of Software Development.
Software Quality Professional, Vol. 3, no. 4, 2001. Available:
http://www.asq.org/pub/sqp/past/vol3_issue4/nonfunctional.html. [Referenced
24.6.2003].

80
Ronkainen, J., Taramaa, J. & Savuoja, A. 2002. Characteristics of Process
Improvement of Hardware-Related SW. In Oivo, M. & Komi-Sirviö, S. (eds.)
2002, Product Focused Software Process Improvement. 4th International
Conference PROFES 2002, LNCS 2559, Springer-Verlag, 247–257.

RTM 2003. Requirements & Traceability Management.

Available: http://www.chipware.com/ [Referenced 15.11.2003].

Ruhe, G., Eberlein, A. & Pfahl D. 2003. Trade-off Analysis for Requirements
Selection. International Journal of Software Engineering and Knowledge
Engineering, Vol. 13, no. 4, 345–366.

Sangiovanni-Vincentelli, A. & Martin, G. 2001. Platform-Based Design and

Software Design Methodology for Embedded Systems. IEEE Design & Test of
Computers, Vol. 18, no. 6, 23–33.

Sawyer, P. & Kotonya, G. 2001. Software Requirements. In: Abran, A., Moore,
J.W., Bourque, P. & Dupuis, R. (eds.) 2001. SWEBOK Trial Version 1.00. Los
Alamitos, USA: IEEE Computer Society Press.

Shull, F., Rus, I. & Basili, V. 2000. Perspective-Based Reading: Techniques for
Improving Requirements Inspections. IEEE Computer. Vol. 33, no. 7, 73–79.

Sommerville, I. & Sawyer, P. 1997. Requirements Engineering. A good practice

guide. Chichester, England: John Wiley & Sons Ltd.

Stankovic, J.A. 1996. Real-Time and Embedded Systems. ACM Computing

Surveys, Vol. 28, no. 1, 205–208.

Stevens, R., Jackson, K., Brook, P. & Arnold, S. 1998. Systems Engineering.
Coping With Complexity. Hertfordshire, England: Prentice Hall.

Tahat, L.H., Vaysburg, B., Korel, B. & Bader, A.J. 2001. Requirement-Based
Automated Black-Box Test Generation. Proceedings of the 25th Annual
International Computer Software and Applications Conference, 489–495.

81
Taramaa, J., Khurana, M., Kuvaja, P., Lehtonen, J., Oivo, M. & Seppänen, V.
1998. Product-based Software Process Improvement for Embedded Systems.
Proceedings of 24th Euromicro Conference, Vol. 2, 905–912.

Thayer, R.H. & Dorfman, M. 1997. Software Requirements Engineering, 2nd

ed. Los Alamitos, USA: IEEE Computer Society Press.

Travassos, G.H., Shull, F., Fredericks, M. & Basili, V.R. 1999. Detecting
Defects in Object Oriented Designs: Using Reading Techniques to Increase
Software Quality. Proceedings of OOPSLA '99, 47–56.

Wiegers, K.E. 1999. Software Requirements. Redmond, USA: Microsoft Press.

Weber, M. & Weisbrod, J. 2002. Requirements Engineering in Automotive

Development - Experiences and Challenges. Proceedings of IEEE Joint
International Conference on Requirements Engineering, 331–340.

Zelkowitz, M.V. & Wallace, D.R. 1998. Experimental Models for Validating
Technology. IEEE Computer, Vol. 31, no. 5, 23–31.

82
Appendix 1: Questionnaire on current state
of requirements implementation practices
Software requirements analysis

Software requirements documentation

• What kind of classification for software requirements is used? (e.g.

functional, interface, quality attributes, design constraints)
• Is there a requirements documentation policy available in the system or software
level? (policy = e.g. procedures, guidelines, templates, standards, etc.)
• If a documentation policy doesn't exist, is it clear how requirements are
documented?

Software requirements' attributes

• What kind of attributes are specified for software requirements? (e.g. id,
priority, stability, rationale, source)
• Are there some requirements attributes missing from the organisation's
current requirement specifications?
• If a priority attribute for a requirement is defined, who are involved in the
prioritisation of requirements (e.g. customers, mangers, developers, etc.)?

Software requirements specification's verification

• Are software requirements specifications reviewed? If yes, how (e.g.
informal review, formal inspection)?
• If specifications are reviewed, who are involved in the reviews (e.g.
managers, designers, coders, testers)?
• Are quality attributes (i.e. non-functional requirements such as reliability,
usability, safety, maintainability) in the SRS verified? If yes, how?

1/1
Software design
Software designers' input for software requirements development
• Are software designers involved in development of software requirements?
If yes, how?

Software architecture's evaluation against software requirements

• Is requirements trade-off analysis made during the architectural design? If
yes, how?
• Is the software architecture evaluated against software requirements during
the design? If yes, how?
• Is the architectural design reviewed? If yes, is the implementation of
software requirements (i.e. how the requirements are actually realised) in the
architecture evaluated?
• If requirements implementation is evaluated in the review, how (e.g.
informal review, formal inspection)?

Detailed designs' evaluation against software requirements

• Are detailed designs evaluated against software requirements during the
design? If yes, how? If not, how is it known that requirements are correctly
realised in the detailed designs?
• Are detailed designs reviewed? If yes, how (e.g. informal review, formal
inspection) and is the requirements implementation evaluated?

Software coding
Software code's evaluation against software requirements
• Is code reviewed? If yes, how (e.g. informal review, formal inspection) and
is the requirements implementation evaluated?
• Is absorption of quality attributes into the code verified? If yes, how (e.g.
how is maintainability, modifiability, portability, efficiency of code
checked)?

1/2
Software testing
Test case development
• When begins the development of test cases for software level testing?
• Are testers included in software requirements development?
• Is unfeasibility of certain requirements found while developing test cases?

Validation criteria for requirements in the SRS

• Are there specific validation criteria for requirements in the SRS? If not, are
there problems testing certain requirements?
• Are there problems testing quality attributes? If yes, with what kind of
quality attributes?

Evaluation of test cases and plans against software requirements

• Is the consistency between test cases and software requirements ensured? If
yes, how? If not, are there problems with indistinct test cases?
• Is the coverage of all requirements within test cases ensured? If yes, how? If
not, are there problems later with untested requirements?

Requirements change management

Requirements change management
• Are there problems in development activities (i.e. requirements analysis,
design, coding and testing) with changing requirements?
• What kind of requirements change management processes, methods, tools
are in use? How are they related to general change management?
• Are software requirements baselined? If not, are there problems with
unstable requirements?
• Is a Change Control Board, Software Configuration Management, or test
case management involved in requirements change management? If yes,
how?

1/3
Requirements change into software requirements specification
• Is there a specific procedure for implementing a change into the SRS that
comes from the system or hardware level? If not, are there problems related
to implementing a change?
• Is there a specific procedure for implementing a change into the SRS that
comes from a software development activity? If not, are there problems
related to implementing a change?

Requirements changes impact to designs, code, and test cases

• Is there a change procedure for handling requirements change into software
designs, code, and test cases? If not, are there problems implementing
change?
• Is there a procedure for impact analysis of requirements change ? (e.g.
process, guidelines, templates, etc. for analysing which components are
affected and how much a change costs)
• To which work products is the impact of a requirement change analysed?
(e.g. to architecture, detailed designs, code, test cases)
• Are there problems while performing impact analysis? If yes, what kind of
problems?

Requirements tracing
• Are requirements traced from system level to software requirements? If yes,
how (tools, matrices)?
• from software requirements to architectures? If yes, how?
• from software requirements to detailed designs? If yes, how?
• from software requirements to code? If yes, how?
• from software requirements to test cases? If yes, how?

• If requirements are traced, is there a requirements traceability policy in software

level? (the policy may inform developers what to trace, why, and how)
• If requirements are traced, do developers have "motivation" to trace? (i.e. do
they know why tracing is needed, and what are the benefits)

1/4
Unrealised requirements
• Is there a procedure for handling unrealised requirements in development
activities? If not, how are unrealised requirements handled?
• Are change management and CCB involved in a clarification of an
unrealised requirement?

1/5
 Appendix 2: Requirements implementation
practices
The following list summarises requirements implementation practices and
related sample methods, techniques, and tools referenced to in this study.

Development activity Practice Sample methods, techniques,

and tools
Software requirements Problem analysis and Structured analysis, object-
analysis product description oriented analysis, formal
methods, data flow diagrams,
state-transition diagrams, use
cases (Parviainen et al. 2003)
Define attributes Priority (QFD, Wiegers
(1999)), Attributes (Sawyer
& Kotonya 2001)
Verification Reviews, Formal inspection,
Perspective-Based Reading
(Shull et al. 2000)
Software design Requirements trade-off ATAM, SBAR, AQA
analysis (MOOSE 2002a)
Consistency check CBSP (Grünbacher et al.
2001), GRL & UCM (Liu &
Yu 2001), SCR & PDL
(Chechik & Gannon 2001),
Inspection (Travassos et al.
1999)
Quality attribute Prometheus (Punter et al.
verification 2002)
Coding Consistency check Instrumented code (Robinson
2002), Analyzer (Chechik &
Gannon 1995)
Quality attribute Phased inspection (Knight &
verification Myers 1993), QAC/++
(2003), Logiscope (2003),
Prometheus (Punter et al.
2002)
Testing Consistency check Automatic Test Scenario
Generation (Gupta et al.
2001), Requirement-based
Automatic Black-Box

2/1
 Testing (Tahat et al. 2001)
Throughout Change management Olsen's change management
development model, Spiral-like change
management model, and
Generic change management
process model (Parviainen et
al. 2003)
Requirements Rational RequisitePro
management (2003), DOORS (2003)
Tracing Matrices (Parviainen et al
2003), DOORS (2003), RTM
(2003)
Impact analysis Quality Factor Deployment,
requirements inspections,
viewpoint analysis, trade-off
analysis (Lam &
Shankararaman 1999),
Wiegers' templates (Wiegers
1999)

2/2
Published by Series title, number and
report code of publication

VTT Publications 526

VTT–PUBS–526
Author(s)
Jäälinoja, Juho
Title
Requirements implementation in embedded software
development
Abstract
Development of correct requirements at the beginning of a software project is considered an
important precondition for successful software development. Moreover, implementing these
requirements correctly during the software development is arguably just as important. Rigorous
implementation of requirements in embedded software development is especially critical, since
requirements affect both software and hardware. The goal of this research is to identify elements
for effective requirements implementation in embedded software development.
A conceptual-theoretical research approach is applied to analyse previous research on requirements
implementation and to construct a new theory which integrates requirements implementation
related elements into a holistic framework. These elements include requirements implementation
processes, methods, and roles. The developed framework describes relations among these elements
and furthermore their relation to software development activities. The framework can be used as a
basis for improving software development areas that are related to requirements implementation.
To validate the feasibility of the developed framework, two case studies were carried out within
embedded software development organisations. The validation was conducted by making a current
state analysis and by suggesting improvements based on the developed requirements
implementation framework. The results from the case studies indicated that the framework was a
useful foundation for improving the organisations' requirements implementation practices.
Keywords
software process improvement, software requirements analysis, embedded systems
Activity unit
VTT Electronics, Kaitoväylä 1, P.O. Box 1100, FIN-90571 OULU, Finland
ISBN Project number
951–38–6370-0 (URL:http://www.vtt.fi/inf/pdf/ ) E2SU00054
Date Language Pages Price
April 2004 English 82 p. + app. 7 p.
Name of project Commissioned by
MOOSE (Software engineering methodologies for
embedded systems)
Series title and ISSN Sold by
VTT Publications VTT Information Service
1455–0849 (URL: http://www.vtt.fi/inf/pdf/) P.O.Box 2000, FIN–02044 VTT, Finland
Phone internat. +358 9 456 4404
Fax +358 9 456 4374
 ESPOO 2004 VTT PUBLICATIONS 526

VTT PUBLICATIONS 526

VTT PUBLICATIONS
509 Sallinen, Mikko. Modelling and estimation of spatial relationships in sensor-
based robot workcells. 2003. 218 p.
510 Kauppi, Ilkka. Intermediate Language for Mobile Robots. A link between the
high-level planner and low-level services in robots. 2003. 143 p.
Juho Jäälinoja
511 Mäntyjärvi, Jani. Sensor-based context recognition for mobile applications.
2003. 118 p. + app. 60 p.
512 Kauppi, Tarja. Performance analysis at the software architectural level. 2003.
78 p.
Requirements implementation in embedded
software development

Requirements implementation in embedded software development

513 Uosukainen, Seppo. Turbulences as sound sources. 2003. 42 p.
514 Koskela, Juha. Software configuration management in agile methods. 2003.
54 p.
516 Määttä, Timo. Virtual environments in machinery safety analysis. 2003. 170
p. + app. 16 p.
515 Palviainen, Marko & Laakko, Timo. mPlaton - Browsing and development
platform of mobile applications. 2003. 98 p.
517 Forsén, Holger & Tarvainen, Veikko. Sahatavaran jatkojalostuksen asettamat
vaatimukset kuivauslaadulle ja eri tuotteille sopivat kuivausmenetelmät.
2003. 69 s. + liitt. 9 s.
518 Lappalainen, Jari T. J. Paperin- ja kartonginvalmistusprosessien mallinnus ja
dynaaminen reaaliaikainen simulointi. 2004. 144 s.
519 Pakkala, Daniel. Lightweight distributed service platform for adaptive mobile
services. 2004. 145 p. + app. 13 p.
520 Palonen, Hetti. Role of lignin in the enzymatic hydrolysis of lignocellulose.
2004. 80 p. + app. 62 p.
521 Mangs, Johan. On the fire dynamics of vehicles and electrical equipment.
2004. 62 p. + app. 101 p.
522 Jokinen, Tommi. Novel ways of using Nd:YAG laser for welding thick section
austenitic stainless steel. 2004. 120 p. + app. 12 p.
523 Soininen, Juha-Pekka. Architecture design methods for application domain-
specific integrated computer systems. 2004. 118 p. + app. 51 p.
525 Mäntyniemi, Annukka, Pikkarainen, Minna & Taulavuori, Anne. A Frame-
work for Off-The-Shelf Software Component Development and Maintenance
Processes. 2004. 127 p.
526 Jäälinoja, Juho. Requirements implementation in embedded software devel-
opment. 2004. 82 p. + app. 7 p.

Tätä julkaisua myy Denna publikation säljs av This publication is available from
Juho Jäälinoja
VTT TIETOPALVELU VTT INFORMATIONSTJÄNST VTT INFORMATION SERVICE
PL 2000 PB 2000 P.O.Box 2000
02044 VTT 02044 VTT FIN–02044 VTT, Finland
Puh. (09) 456 4404 Tel. (09) 456 4404 Phone internat. +358 9 456 4404
Faksi (09) 456 4374 Fax (09) 456 4374 Fax +358 9 456 4374

ISBN 951–38–6370–0 (URL: http://www.vtt.fi/inf/pdf/)

ISSN 1455–0849 (URL: http://www.vtt.fi/inf/pdf/)