Requirements Reuse for Improving Information Systems Security:

A Practitioner's Approach
Ambrosio Toval, Joaqun Nicols, Begoa Moros, Fernando Garca

Software Engineering Research Group. Departamento de Informtica y Sistemas

Universidad de Murcia. Campus de Espinardo. 30071. Murcia (Spain) {atoval, jnr, bmoros}@um.es
Innosec (Innovative Secure Comunications, Inc.). Paseo de la Castellana, 118

28046 Madrid (Spain) [email protected], www.innosec.es

ABSTRACT
Information systems security issues have usually been considered only after the system has been
developed completely, and rarely during its design, coding, testing or deployment. However, the
advisability of considering security from the very beginning of the system development has
recently begun to be appreciated, and in particular in the system requirements specification phase.
We present a practical method to elicit and specify the system and software requirements,
including a repository containing reusable requirements, a spiral process model, and a set of
requirements documents templates. In this paper, this method is focused on the security of
information systems and, thus, the reusable requirements repository contains all the requirements
taken from MAGERIT, the Spanish public administration risk analysis and management method,
which conforms to ISO 15408, Common Criteria Framework. Any information system including
these security requirements must therefore pass a risk analysis and management study performed
with MAGERIT. The requirements specification templates are hierarchically structured and are
based on IEEE standards. Finally, we show a case study in a system of our regional administration
aimed at managing state subsidies.

Keywords: Requirements Engineering; Requirements Reuse; Security; Common Criteria

Framework; Risk Analysis and Management Methods.

1 Introduction
There has recently been an increasing interest in information systems security issues. For
instance, PITAC (President's Information Technology Advisory Committee) [1] has
stated the need for a scalable information infrastructure, that is, for techniques which
ensure that the United States information infrastructure -including communications
systems, the Internet, large data repositories, and other emerging systems- is reliable and
secure and can grow smoothly to accommodate the massive numbers of new users and
applications requiring high bandwidth which are anticipated over the next two decades.
Nowadays, information systems are vulnerable to many threats, such as new
viruses (e.g. worm viruses) that propagate through Internet; the threats brought about by
unacceptable employees use of the Internet resources (such as non-business activities,
accidental or deliberate disclosure of sensitive information, and hacking) [2]; failure to
observe the personal data privacy laws (leading, in our country, to fines of up to
$700,000 or to important administrative sanctions [3, 4]); even strikes or loss of key
personnel are threats that information systems need to be able to cope with. In addition,
the general acceptance of e-commerce and the digital signature to perform
administrative and commercial transactions means that the security of information
systems needs to be ever more reliable. A security breakdown can result in very serious
problems for an organization: for example, a recent survey on 1,000 UK organizations
[5] shows that the occurrence of a security failure in a business with no contingency plan
leads to its shutdown in 80% of the cases, and 60% of UK businesses have suffered an
important security breach in the last two years.
A large number of methods and regulations concerning the security of
organizations have appeared in response to such a situation. Of particular importance are
ISO 15408, Common Criteria Framework (CCF) [6], and the following national risk

analysis and management methods: CRAMM in the UK (CCTA Carmarthenshire

College Of Technology and Art Risk Analysis and Management Method) [7], MARION
in France (Mthode dAnalyse de Risques Informatiques et dOptimisation par Niveau)
[8], and MAGERIT (Metodologa de Anlisis y GEstin de RIesgos del MinisTerio de
Administraciones Pblicas) [9], which is the Spanish Public Administration's adaptation
of CCF.
In the information systems development field, requirements form the foundation
for the rest of the software development process, since building a high-quality
requirements specification is essential to ensure that the product satisfies the users
needs [10-12]. However, drawing up a specification of quality requirements is a difficult
task. According to the IEEE 830-1998 standard [13], a requirement of quality is that it
be correct, unambiguous, complete, consistent, ranked for importance and/or stability,
verifiable, modifiable, and traceable. We agree with Mili et al. [14] that there exist few
alternatives but software reuse as the (only) realistic approach to bring about the gains of
productivity and quality that the software industry needs. It has been agreed (see, for
example, [12, 15]) that the benefits of reuse are greater when the abstraction level is
increased and not only code, but also designs and specifications, are reused.
The usual practice during the early steps in information systems development has
been to pay attention to such aspects as reliability, availability or integrity, while
security issues have usually been considered only once the system has been deployed, or
at best, during the system design, coding or testing [9, 16].
In this paper, we present a proposal to consider information systems security
beginning with the Requirements Engineering (RE) process. The approach is based on
the reuse of security requirements which are compatible with MAGERIT (and, therefore,
also with a CCF subset) and which are collected in a reusable requirements repository.

This approach complements SIREN (SImple REuse of software requiremeNts), a general

purpose RE method based on requirements reuse (see Figure 1) that we are currently
defining. As this paper explains, the inclusion of requirements from the repository
controls risk levels within the information system to be built, so that the information
system will fulfill the demands of risk analysis performed with MAGERIT. How this
approach has been applied to a real case study concerning an information system in our
regional government is shown in this paper.
RE standards and
recommended
practices
IEEE 830-1998
IEEE 1233-1998
...

Guidelines
and
Structure

Requirements
specification
documents

IRS

MAGERIT

Reusable
Requirements
Repository
Domain
Analysis

Requirements
Reuse
Requirements
Documentation

SyRS

SyTS

SRS

STS

Requirements
Elicitation,
Analysis
and
Negotiation

Requirements
Engineer

Requirements
Validation
Stakeholders

Figure 1. A brief summary of our proposal for requirements reuse, particularized to the security field.
The structure of the requirements specification documents is explained in section 4.1.

The structure of this paper is as follows: Section 2 justifies the interest of the problem
addressed by our proposal; Section 3 includes a brief description of MAGERIT, the
structure of which determines the structure of the security requirements in the repository;
Section 4 summarizes our RE process SIREN, and Section 5 describes the case study.
Finally, Section 6 shows the conclusions and outlines further research.

2 Reasons behind the study

This research is based on the lessons learned during a risk analysis and management
project developed with MAGERIT in our regional government (explained in Section 5).

Our research offers a combination of RE, reuse, and information systems security in
order to improve information systems quality and productivity.
2.1

RE and Security

The essential reason behind this proposal is the need to include security explicitly in
information systems development, as of the RE process, and thus improve several
information systems quality attributes, namely security, safety, reliability, and
robustness. Sommerville [12] claims that a high percentage of system malfunctions are
the result of errors in specification rather than design. For example, the specification
may be incomplete in that it does not describe the behavior of the system required in
some critical situations. In a study of errors in embedded systems, Lutz [17] concludes
that "(...) difficulties with requirements are the key root cause of the safety-related
software errors which have persisted until integration and system testing".
Although the widespread use of Internet has given rise to very active,
multidisciplinary research concerning security issues, and researchers in RE are
beginning to focus on e-commerce and other Internet issues, we agree with Antn and
Earp [16] that there remains a need to apply proven RE methods in this field. Some
interesting approaches are the following:

The integration of SSADM (Structured Systems Analysis and Design Method) [18]
with CRAMM [7], and Mtrica 3 [19] with MAGERIT (Mtrica 3 is the standard
development method of the Spanish public administration and is similar to SSADM).
These approaches relate standard information systems development methods
(including requirements specification) to standard risk analysis and management
methods, and thus address the entire system-development life cycle. They do not,
however, provide a reusable requirements repository to support reuse of security
safeguards during the RE process. This criticism also concerns PFIRES (Policy

Framework for Interpreting Risk in eCommerce Security) [20], a non-standard

framework for risk analysis in e-commerce which addresses the entire systemdevelopment life cycle. The plan phase of PFIRES includes a requirements definition
step, but it does not provide specific guidance on how to translate the security
recommendations into requirements (safeguards), as does, for example, the
integration between Mtrica 3 and MAGERIT.

First Chung [21] and later Antn and Earp [16] focused their previously proposed
goal-based RE frameworks on the security field (with the latter specifically
addressing e-commerce security issues and built on the PFIRES approach). We think
that these approaches can be complementary to ours: as we explain below in this
paper, we use MAGERIT in the first place to perform a risk analysis, and then we
manage risk by reusing MAGERIT-compatible security requirements. Analogously,
we could first specify the goals of the system to be developed by means of one of the
goal-based frameworks mentioned before, and then operationalize the goals by
making use of our security requirements, and thus reuse generic security design
knowledge from MAGERIT. Like this, we would be able to specify the relationships
between the security needs and the other functional and non-functional requirements.

Jones et al. [22] are carrying out a project aimed at defining a framework for trust
requirements in e-business. This framework contains three major elements:
stakeholders in e-business, a list of common e-business process models, and a
conceptual model of the elements involved in an e-business transaction. These
authors, moreover, seek to develop a list of generic requirements for e-business to be
adapted to specific projects by means of the framework. The purpose of this work is
very similar to the research that we present in this paper, but the work focuses
specifically on e-business, while our work is focused on information systems

security in general. Their research is still at an early stage, and they do not present a
list of requirements, but a list of categories of generic requirements.
Security standards do not adequately address the RE process either. We consider that a
significant part of the study of Fenton and Neil [23] on lacks of safety related standards
is also applicable to security standards (in particular, to ISO 15408 Common Criteria
Framework). In particular, we agree with Fenton and Neil that: i) the requirements
imposed by the standards are imprecise; and ii) the standards are too complex and
difficult to use. In fact, Fenton and Neil show a process for improving safety-related
standards through the improvement in the structure and writing of the requirements that
the standards impose. These authors, moreover, claim that safety standards are
conceived to evaluate systems already created, and they do not consider the problem of
providing security for new systems.
2.2

RE process improvement

Our proposal also addresses the improvement of information systems quality through the
improvement of the quality of the RE process. In this section we deal with the
improvement of the quality of the RE process by means of the main processes life cycle
standards and quality standards. To this extent, neither life cycle processes standards
(such as ISO 12207 [24]) nor quality standards (such as ISO 9000 [25]) provide precise
guidelines for identifying problems and planning improvements in the life cycle
processes of systems and software (in particular, ISO 9000 does not include any section
specifically devoted to RE [26]).
In contrast, capability maturity models, such as CMM [27] and ISO
15504/SPICE (Software Process Improvement and Capability dEtermination) [28], do
provide guidelines to improve the quality of the life cycle processes. However, once
again, they do not cover the RE process [26]. This circumstance led Sommerville and

Sawyer to propose a Requirements Engineering Process Maturity (REPM) model [26],

which defined a set of good practices in RE, and allowed organizations to be assessed at
three levels, namely 1) Initial; 2) Repeatable; and 3) Defined. Level 1 organizations do
not have a defined RE process, while level 2 organizations have defined standards for
requirements documents and requirements descriptions and have introduced policies and
procedures for requirements management. These two levels correspond to CMM levels 1
and 2. Finally, REPM level 3 is assigned to organizations that have a defined RE process
model based on good practices and techniques, with an active process improvement
program in place and which can objectively estimate new tools and techniques. This
third level corresponds to CMM levels 3, 4, and 5.
REPM is based on the incremental adoption of an RE good practice guide [26].
This guide, on which our proposal SIREN is based, contains a spiral process model (see
section 4.3) and ten good practice guidelines (see Table 1).
Define a standard document structure
Make the document easy to change
Uniquely identify each requirement
Define policies for requirements management
Define standard templates for requirements description
Use language simply, consistently and concisely
Organize formal requirements inspections
Define validation checklists
Use checklists for requirements analysis
Plan for conflicts and conflict resolution
Table 1 Sommerville and Sawyer's top ten guidelines.

3 MAGERIT
MAGERIT [9] is the information systems risk analysis and management method of the
Spanish public administration, which is compatible with ISO 15408, Common Criteria
Framework. In this paper we show how we have translated the security measures stated
in MAGERIT into reusable security requirements. MAGERIT is, therefore, the source of
our reusable security requirements repository. MAGERIT, moreover, specifies how risk

analysis has to be performed before selecting the required security requirements (see
Section 4.3). This section summarizes the basic elements of MAGERIT beginning with a
brief description of the evolution of the risk analysis and management methods.
Baskerville [29] studies the evolution of information systems security design
methods. The first generation of methods for analyzing risk in information systems
appeared in the United States in the early 1970s, based on checklists. From the mid
1980s, there appeared a second generation of more formalized methods headed by the
British CRAMM. In the late 1990s, there began to appear a third generation of methods
more linked to information systems development methods and closer to security
legislation and regulations, that is, 3rd generation security methods take into account the
personal data protection laws as well as some other regulations in the security field. The
new version of CRAMM (integrated with SSADM) and the first version of MAGERIT
(integrated with Mtrica) pertain to this generation. Organizational problems are
considered in the third generation and so security is reconciled with the functionality of
the system, since the security aspects are taken into account from the first stages of the
development process [29].
MAGERIT is, in fact, only applied to one of the steps in information systems
security management: that of risk analysis and management. Starting from the
objectives, strategy and policy of security in the existing information systems in the
organization, MAGERIT is applied in order i) to study the risks that affect each
information system and its environment, and ii) to propose the countermeasures
(safeguards) that should be adopted to register, prevent, avoid, reduce and control the
risks evaluated. These countermeasures are collected in an information systems security
plan which is then put into practice, and followed up during the maintenance stage. The

application of MAGERIT is iterative, so that new risk analyses are periodically

performed, and hence new countermeasures are proposed during maintenance.
The risk analysis and management model of MAGERIT includes:

the submodel of elements, providing the basic entities related to the information
system risk analysis: assets, threats, vulnerabilities, effects, risks, and
countermeasures;

the submodel of processes, describing the stages in the security project that is to
be developed: planning, risk analysis, risk management, and recommendation of
countermeasures.

Risk analysis with MAGERIT involves the following major steps: i) identification of the
assets of the organization, which are the resources of the information system, and which
may either directly belong to the information system or just to the information system
environment; ii) study of the vulnerabilities of these assets and the threats to them; iii)
estimation of the risk related to these assets, based on the threats and vulnerabilities
associated (threats are qualified by the likelihood of their occurring); iv) proposal of the
countermeasures managing the risk. Therefore, countermeasures manage threats and are
transitively linked to assets.
As indicated at the beginning of this section, we have translated the MAGERIT
countermeasures into security requirements. Security requirements are thus linked to
assets. Once risk analysis has been performed and we know how the assets are menaced,
it is useful for the structure of the security requirements in the repository (Section 4.2) to
conform to the structure of the assets of MAGERIT in order to find the suitable security
requirements (countermeasures). It is, therefore, of interest to briefly outline the
organization of the assets in MAGERIT (Figure 2).

10

Other
assets
Organizations
functions

Information

Information system

Environment

Figure 2. MAGERIT assets hierarchy.

As Figure 2 shows, MAGERIT organizes assets in five layers:

The information system environment layer includes the facilities containing

the information system, such as furniture and supplies.

The information system layer includes the information related to the software,
hardware, and personnel, which make up the information system.

The information layer includes the information managed by the information

system: applications data and metadata (such as data structures and data
dictionaries).

The organizations functions layer justifies the usefulness of the information

system, by providing it with a purpose; it describes the assets and services
produced by the information system, and the users of the information system
services.

Finally, the other assets layer describes assets that are usually intangible, and
do not fit into the lower layers, as, for instance, the organization credibility or
an individuals privacy.

Each layer can be refined into blocks of homogeneous assets. Besides, each block
can be further divided into sub-blocks. For instance, Figure 3 shows the structure of
the information system layer. The initials (IS, ISHW, ISSW, etc.) in this figure are

11

used for maintaining the structure of the MAGERIT asset hierarchy in the security
requirements repository, as explained in Section 4.2.
IS.HW.MS
Magnetic
Storage

A requirements layer
IS

IS.HW.LWA
IS.HW
Hardware

Servers

Software

Peripherals

IS.SW.IS

IS.SW.IS.S
Security
Services

Personnel

IS.SW

IS.HW.P

Internet
Services

IS.P

LAYER 2
Information
System

LAN/WAN
Architecture
IS.HW.S

A requirements block

IS.SW.CSA
IS.SW.O
Control and
Office
Systems
Software
Applications

IS.SW.E
External
Software

SI.SW.ADM
Applications
for tailored
development

IS.SW.E.AV
Antiviruses

IS.SW.IS.CS
Communication
Services

IS.SW.OS
Operating
Systems

IS.SW.OS.AC
Access Control
IS.SW.OS.RM

IS.SW.IS.S.F
Firewall
IS.SW.IS.S.P
Proxy

Rights management

IS.SW.OS.AL
Activity logs

Some more detailed requirements blocks

Figure 3. An extract of the MAGERIT information system layer.

4 The SIREN approach to requirements reuse

Our proposal, called SIREN (SImple REuse of software requiremeNts), is a method for
RE based on requirements reuse. The purpose of development with requirements reuse is
to identify descriptions of systems that could be used (either totally or partially) with a
minimal number of modifications, thus reducing the total effort of development [15]. In
our view requirements reuse can produce more benefit than only design or code reuse
because traceability relationships (see Section 4.2) can be established between the highlevel requirements of the system and the architecture and implementation which are built
from them. Thus, if a system requirement is reused, the subsequent development process
can be speeded up. Although a lot of more complex techniques for reusing requirements
exist (see the Cybulsky's classification in [15]), our SIREN proposal for requirements
reuse emphasizes simplicity and it is therefore more practical.

12

SIREN also conforms to the most well-known Software Engineering standards

for requirements specification (Section 4.1). Requirements have a textual format, but can
include any kind of objects as complementary information - for instance, tables or
schemas of any type.
SIREN encompasses a process model, some guidelines, techniques and tools. The
guidelines that SIREN provides consist of a hierarchy of requirements specification
documents together with the templates for each document (Section 4.1). These serve to
structure a reusable requirements repository (Section 4.2). Finally, we will present the
SIREN process model (Section 4.3) and a discussion on the tools used to support the
process (Section 4.4).
4.1

Requirements documents hierarchy

The SIREN requirements documents hierarchy that we propose is depicted in Figure 4.

In accordance with Gabb [30] we consider that each document correspond to a different
specification level and, therefore, it has different objectives and users.

IRS
Interface Requirements
Specification
(IEEE Std. 830)

SyRS
System Requirements
Specification
(IEEE Std. 1233;
IEEE Std. 12207.1)

SyTS
System Test Specification

SRS
Software Requirements
Specification
(IEEE Std. 830)

STS
Software Test Specification

Figure 4. SIREN requirements documents hierarchy.

SyRS. System Requirements Specification

The SyRS describes the function and performance of the whole computer-based system,
together with the constraints that will govern its development and the information (data
and control) that is the input and the output of the system. It serves as the foundation for

13

hardware engineering, software engineering, database engineering and human

engineering [31].
According to IEEE 12207.0 [24], the SyRS should include: i) functions and
capabilities of the system, ii) business, organizational, and user requirements, iii) safety,
security, and privacy protection requirements, iv) human-factors engineering
requirements, v) operations and maintenance requirements and vi) design constraints.
From these features, IEEE 12207.1 [32] details the specific contents of the SyRS
template referencing the IEEE Std 1233 [33], Guide for Developing System
Requirements Specifications, where some guidance on specifying the system
requirements and an SyRS outline can be found. This outline is the basis of the SIREN
SyRS template.
SRS. Software Specification Requirements
In the SRS, the functions and performance allocated to software as part of system
engineering are refined by establishing a complete information description, a detailed
functional description, a representation of system behavior, an indication of performance
requirements and design constraints [31].
The SRS includes requirements on the software functionality, external interfaces,
performance, design constraints, and software attributes (portability, maintenance,
security, availability, and reliability). Most software requirements are directly derived
from the system requirements; hence the hierarchical vision presented in Figure 4. The
SIREN SRS template is based on the IEEE Std 830-1998 [13] and on the VOLERE
template proposed by Robertson and Robertson [11].
SyTS and STS. System and Software Testing Specifications
Any requirement in the SyRS or SRS must be quantifiable in order to be subsequently
validated (see, for instance, Robertson and Robertson [11] or Sommerville [12]). Each

14

requirement in the SyRS and SRS needs to have been linked to a testing criteria (called
fit criteria in [11]) in the SyTS and STS documents, respectively. These testing criteria
specify how to check that the system or software implemented fulfills the requirements
defined.
We believe that SyTS can be very useful in the security field because the security
plan of the organization can be specified directly from the SyTS. This plan will include a
list of security questions (related to personnel, organization, and functions) which may
be then easily checked.
IRS. Interface Requirements Specification
The requirements related to the interfaces between the software elements and between
software and users can be included in the SRS. Nevertheless, in order not to produce
overlong documents, it is sometimes useful to collect all this information in a separate
document, called IRS. Therefore, traceability relationships (Section 4.2) need to be
established between the SyRS, the SRS modules and the interfaces described. The IRS
template has the same structure as the SRS section for specifying interfaces.
Documents hierarchy definition and evolution
These five types of documents do not have to be delivered in each project. The project
manager (together with the analysts) decides which hierarchy to use, depending on the
project dimension and the complexity of the system to be developed: i) The System
Requirements Specification (SyRS) and the System Test Specification (SyTS) are
unique, but they are not necessary when the problem is simple and when it is clear from
the beginning that all the requirements will be supported by software. ii) The Software
Requirements Specification (SRS) is compulsory, but can be divided into sub-documents
(i.e. one SRS for each subsystem or module). iii) By contrast, the Interface
Requirements Specification (IRS) is optional and is used when the interface

15

requirements are numerous in order to avoid producing a very long SRS. If this is not the
case, they will be included in the SRS. iv) Finally, there should be a Software Test
Specification (STS) for each SRS in the hierarchy.
The definition of the hierarchy of documents is performed iteratively during the
RE process. This hierarchy usually begins with one SyRS and one SyTS (see section
4.1), along with one SRS (and the corresponding STS). Subsequently, the hierarchy of
documents will vary gradually during the RE process: for instance, the SRS could be
split into several sub-documents, one for each subsystem identified.
4.2

Contents of the Requirements Repository

The repository contains requirements from specific domains and profiles. SIREN
domains consist of requirements belonging to a specific application field, such as
accounting or finance. A SIREN profile, on the other hand, consists of a homogeneous
set of requirements that can be applied to a variety of domains, such as information
systems security (the one presented in this paper), and the personal data privacy law
[34]. The structure of the domains and profiles of the repository (see Figure 5) is
determined by the previous requirements specification documents hierarchy.
There are two types of requirements in the repository:

parameterized: this kind of requirement contains some parts that have to be adapted
to the application being developed at the time. For example:
SRS.3.5.3.1.S.301 The security manager shall check the users identifiers every
[time in months] to detect which ones have not been used in the last [time in months].

If this requirement is chosen, the parameterized part will be instantiated, that is, the
information in brackets will be replaced by a specific value according to the current
project. The requirement R4 in Figure 6 represents another example.

Although some requirements management tools, such as RequisitePro, do not allow the inclusion of
points in the labels, they are used in the example to improve legibility.

16

non-parameterized: requirements that could be applied directly to any project

concerning the profiles and/or domains in the repository. The requirements in Figure
6 (except for R4) and in Figure 7 are examples of non-parameterized requirements.

According to the recommendations of the IEEE 1233 standard [33], each requirement
has the following attributes: identification (unique), priority, criticality, viability, risk,
source, and others which depend on the project. Although the attributes are not shown in
Figure 6 and Figure 7, they are included in the repository.
In our approach, the requirements are labeled in such a way that both their profile
or domain, and their location in the document are shown at the same time. For instance,
in Figure 5, label SRS.3.5.3.1.S.23 refers to security requirement 23 (S23) in the
Software Requirements Specification document (SRS), and it is located in section
3.5.3.1, which deals with the system attributes (section 3.5) related to security (3.5.3),
and in particular to confidentiality (3.5.3.1).

SRS

3. Specific Requirements

3.5 Software Systems Requirements

3.5.3.Security
3.5.3.1 Confidentiality
SRS3531S23 The user ...
SRS3531L12 The audit ...
SRS3531BD9 The database ...

Reusable Requirements Repository

SRS

3. Specific Requirements

3.5 Software System Attributes

3.5.3.Security
3.5.3.1 Confidentiality
SRS3531S23 The user ...

Security Profile

SRS

3. Specific Requirements

3.5 Software System Attributes

3.5.3.Security
3.5.3.1 Confidentiality
SRS3531L12 The audit ...

Personal Data Protection Profile

SRS

3. Specific Requirements

3.5 Software System Attributes

3.5.3.Security
3.5.3.1 Confidentiality
SRS3531BD9 The database ...

... Profile

Figure 5 Reusable requirements repository composed of profiles and domains. Each profile or domain is a view of
the global repository. In the figure, only the SRS document is shown.

Two additional attributes have been defined in the security profile:

17

MAGERIT-location, whose values are the initials that specify the position of
the security requirement within the MAGERIT asset hierarchy (see Figure 3).
For example, the initials IS.SW.IS.S.F (Firewall sub-block in the information
system layer, which can be found in Figure 3) are the value of this attribute
for the requirements in Figure 7. In this way, the security profile in the
requirements repository may be sorted following two criteria: the structure of
the requirements specification documents and the structure of the MAGERIT
asset hierarchy.

Obligement level, providing the analyst with information about the

requirements that have to be included. The possible values for this attribute in
the security profile are: compulsory, recommendable and optional.

The traceability relationships between the requirements of the repository also have to be
collected (see Ramesh and Jarke [35] for a general discussion on traceability). In the
current SIREN model, a trace can be established between requirements belonging to the
same or different documents (see Figure 6). These traces can imply inclusive or
exclusive dependencies:

An inclusive trace between two requirements A and B means that to satisfy A, B also
needs to be satisfied and, therefore, the reuse of A implies the reuse of B. Figure 6
shows an example of an inclusive trace between requirements in the SyRS and SRS.
We can observe that for

R2

to be achieved,

R1

and

R3

(and others which are not

shown) have to be accomplished, and in order to fulfil R3, R4 also has to be fulfilled.
Hence, R2 also depends, transitively, on R4.

18

SyRS (System Requirements Specification)

3. System Capabilities. Conditions and constraints.
3.1. Physical.
3.1.1. Construction.
R1 (SYRS311S68) Documents and diskettes shall be kept under lock and key
when they are not being used and out of office hours.
3.6. Policy and Regulation.
R2 (SYRS36S26) The authorized user of the information system shall know
their duties related to the access controls and the information under their
responsibility. Thus, three conditions are stated:
a) The authorized users have to use their password properly.
b) The authorized users can not for a moment disregard the
information that they manage.
c) The authorized users have to follow the security measures enforced
to avoid unauthorized accesses to the information managed by
them.
Trace-to: R1, R3 (...)
SRS (Software Requirements Specification)
3. Specific Requirements.
3.5. Software Attributes.
3.5.3. Security.
3.5.3.1. Confidentiality.
R3 (SRS3531S1) The operating systems used shall provide password
mechanisms to control and/or limit the access of the users.
Trace-to: R4
R4 (SRS3531S14) The operating system used shall provide programs to verify
the quality of the passwords in the Access Control System. A password is said
to be of quality if it fulfils at least these three features:
a) The minimum number of characters is [n, n >= 6].
b) It has at least [n, n>=1] numeric characters and [m, m>=1]
alphanumeric characters.
c) It shall be changed every [time in days] by general users and every
[time in days] by users with privileges.
Figure 6. Examples of system (SyRS) and software (SRS) requirements.
Labels R1-R4 are only used as abbreviations.

An exclusive trace between two requirements means that they are mutually
alternative. An example is shown in Figure 7. Since it is not possible to configure the
same firewall in two different ways at the same time, only one of the requirements
will be selected, according to the security needs of the specific project.
SRS.3.4.3.S.5. The firewall configuration will be screened host.
Exclusions: SRS.3.4.3.S.6
SRS.3.4.3.S.6. The firewall configuration will be screened subnet.
Exclusions: SRS.3.4.3.S.5.

Figure 7. Example of exclusive trace between requirements in the SRS. Section 3.4 in the SRS deals with the design
constraints and includes the subsection 3.4.3. Related Applications.

Traceability can be followed forwards or backwards from a given requirement. Later in

the development of the system, forward traces can be used for linking requirements to

19

the rest of development artifacts (analysis, design and implementation models), thus
aiding the evolution and maintainability of the system.
4.3

Process model

According to Kotonya and Sommerville [10], RE basically comprises the phases of i)

requirements elicitation, ii) requirements analysis and negotiation, iii) requirements
documentation and iv) requirements validation. We agree with these two authors that a
spiral life cycle [10] (as shown in Figure 8) is the most suitable way to cope with the RE
process, since it is not possible to obtain a complete set of the requirements in a single
iteration. This model has been adapted in SIREN to take into account requirements reuse
explicitly in the RE process.
Informal
requirements

Reuse of
requirements

Analysis and
Negotiation

Elicitation
Requirements
document and
validation report

Agreed
requirements
Validation

Repository
Improvement

Documentation

Draft
requirements
document

Figure 8. Spiral model for RE adapted to SIREN.

In order to use the security profile properly, the Elicitation phase must start by studying
the threats to the assets of the system. Next, the risk concerning those assets will be
estimated depending on those threats and the vulnerability of the assets. Finally, a set of
requirements from the repository will be selected to establish the countermeasures
required to manage the risk analyzed.
The SIREN process model is shown in Figure 9 in more detail. The requirements
documents templates (presented in Section 4.1) guide the requirements engineers in the

20

Elicitation phase since they encompass all the aspects that must be taken into account to
write the requirements specification. We divide this phase in Figure 8 into two activities:
a) Requirements selection. The approach for reuse consists of providing the analyst
with the specification documents templates filled in with requirements. In the
domains and profiles related to the current project, the analysts together with the rest
of the stakeholders reuse those requirements from the repository that are suitable for
the specific application that is being developed. It should be noticed that when a
requirement is reused, all the requirements related to it by means of an inclusive
dependence will also be selected. In addition, the CARE (Computer-Aided
Requirements Engineering) tool supporting the process should ensure that two
exclusive requirements will not be inserted in the specification.
b) Specific requirements elicitation. In the meetings with the rest of the stakeholders,
the analysts gather the informal requirements of the problem domain as requirements
drafts, which contain the specific knowledge about the system that is going to be
developed. These requirements are not in the repository initially they are specific
for the problem at hand.
The templates with reused requirements and the informal requirements documents are
the input for the Analysis and Negotiation activity, the aim of which is to discover
problems with the system requirements and reach agreements on changes which satisfy
all the stakeholders. This negotiation has to remove redundancy and inconsistency. At
the end of this activity, the Agreed Requirements are identified and then specified in
detail in the Documentation activity. This latter activity leads to some Draft
requirements documents, which will follow the selected documents hierarchy. The draft
requirements are validated by the stakeholders and thus the Validated Requirements
Documents, containing the needs and constraints of the information system, are

21

produced. From these documents, a new spiral in the RE process starts. Eventually, the
Validated Requirements Documents are considered complete enough to go onto the
development process.
Reusable
Repository

Requirements
selection
Security Profile
Data Protection

Filled in
templates
with reused
requirements

SyRS

SyTS

IRS

SRS

STS

...
Empty
templates
IRS

SyRS

SRS

SyTS

STS

Informal
Requirements

Validated
Requirements
Documents

Analysis
and
Negotiation

Specific
Requirements
Elicitation

Repository
Improvement

SyR

SyTS

SRS

STS

Agreed Requirements

IRS
IR

SyTS

SRS

STS

Documentation

Validation

Going on: analysis,

design,
implementation, ...

SyRS

IR

SyR

SyTS

SRS

STS

Draft Requirements
Documents

Stakeholders
Analyst

Figure 9. SIREN approach for requirements reuse.

It is important to point out that we are not proposing a waterfall model. These
specification documents will evolve during the rest of the life cycle; for instance, during
the design, the specification could be enriched with requirements related to the
technological environment or the user manuals (such as organization and appearance).
We agree with Nuseibeh [36] that the RE and architecture design processes are
concurrent and influence each other.
A potentially inherent problem to any requirements specification is premature
design. In particular, security requirements tend to propose design solutions (e.g.
password-based validation in Figure 6) and these may not be appropriate in all

22

circumstances. In our approach, security requirements specify the countermeasures

prescribed by MAGERIT after the risk analysis. Therefore, it is the MAGERIT risk
analysis and management which determines the security mechanism to be used in each
circumstance. In this way, the security of the information system relies on the
MAGERIT risk management and on the proper design and implementation of the
security requirements.
The repository shall not be considered as a final static product, but as an
evolutionary product that has to be enriched with those new requirements found
throughout the development of the current system and which are considered as likely to
be used in forthcoming applications. Furthermore, the requirements already in the
repository could be modified in order to improve their quality. This is the aim of the
Repository Improvement activity in Figure 9.
4.4

Tools

To support the process presented in the previous section, we need tools both for
managing the requirements and for supporting reuse.
There are many commercially available tools for managing requirements, such as
CORE by Vitech Corporation [37], DOORS by Quality Systems & Software Inc (QSS)
[38] and Requisite Pro by Rational [39], but none of them provided us with all the
required functionality. All of these tools share the same problem: neither parameterized
requirements nor exclusive relationships (section 4.2) are supported. In addition, they
lack features for supporting reuse.
It is in such a context that we are considering the implementation of our own tool.
We want to evaluate Sommervilles recommendation [26] that the easiest way to provide
reuse is by means of a documentary database (such as INVESDOC by IECISA [40]), so
that the requirements could be searched by key words, and parameterized and exclusive

23

requirements could be easily implemented. Until now, we have used Requisite Pro
because it is widespread, good value for money, it is integrated with the rest of the
Rational development tools, and it is easy to use.

5 Case study
As we said at the beginning of Section 2, the approach that we present in this paper
stems from the lessons learned during the development of the CARMMA project [41]. In
this project our regional government contracted our research group to develop a risk
analysis

using

MAGERIT

in

the

Regional

Information

Systems

and

Telecommunications Office (RISTO). The staff at RISTO were very interested in the
improvement of the security of their systems since their information must be constantly
available to the other regional government offices, and a huge amount of resources
(equipment, people and money) are involved in RISTO. A few figures will enable the
reader to realize how critical a security breach could be: our regional government has
roughly 22,000 employees and a budget for the year 2000 of nearly $1,275 million,
while RISTO employs 40 people and has a budget of $2.89 million.
The risk analysis took one year and involved five analysts and some fifty
stakeholders. The stakeholders included the following roles: managers and
administrative staff (as final users) and technical staff. From the results of the study we
realized that:

the cost of implementing the countermeasures to reduce the risk of most of

the critical assets identified would have been lower if these assets had been
developed taking into account security issues from the beginning;

although the countermeasures in MAGERIT are linked to threats, instead of

to assets, we think that the implementation of each countermeasure depends

24

on the specific asset threatened and, therefore, the countermeasures should be

linked to assets.
These conclusions lead us to the definition of a SIREN security profile (Section 4.2)
structured in line with the MAGERIT asset hierarchy. This repository helps to make
security issues explicit from the early steps in the system development process.
We considered that it would be very interesting to validate our proposal in a new
case study. We preferred to carry out the new case study in our regional government in
order to make use of the knowledge gained in the previous project and so we would be
working as security consultants for the office selected. Moreover, we wanted to
participate in the requirements specification of a new information system so that we
could define the security requirements from the beginning of the development process.
This new information system would be secure after the proper implementation of its
requirements.
A case study in the Regional Ministry of Labor and Welfare, satisfying the
requirements stated above, was then selected to experiment our approach. The project
dealt specifically with the management of state subsidies awarded to put into effect
activities including the promoting of the business realm and career advice, rural
development plans and occupational safety and health measures.
The work group was formed by three people, one from our research group
(working as analyst) and the other two from the regional government (working as
stakeholders). It is worth pointing out the following two activities carried out by the
work group:
1) The performance of a risk analysis to detect security needs (threats, critical assets
and countermeasures). An asset is critical either because it is very valuable although
it is barely threatened or because although it is of little value it is heavily menaced.

25

The value of an asset is calculated according to the loss that it would cause if it
failed, that is, the value of an asset is an accumulated value as shown in Figure 10.
For example, the estimated value of the power hub in Figure 10 is $35,897 but its
accumulated value is $24,174,348, corresponding to a function of the weighted sum
of the values of the assets that rely on the power hub (not all the assets involved are
shown in Figure 10 for reasons of simplicity). These values can be used to prioritize
the asset (and therefore the countermeasures related to it).
ESTIMATED VALUES
APP:

ACCUMULATED VALUES
SOFTWARE: CUSTOM
BUILT APPLICATION

CUSTOM BUILT APP

APP

$1,282,051
SOFTWARE:
DBMS

ORACLE:

ORACLE DATABASE MANAGEMENT SYSTEM

ORACLE

$51,282
SOFTWARE:
OPERATING SYSTEM

O.S. UNIX:

OPERATING SYSTEM HP-UX

$30,769
SERVERS:

$102,574

SUB-NETWORK:
$0

HARDWARE:
SERVERS AND
LAN/WAN

LOGIC SUB-NETWORK 111.11.111.X

POWER HUB:
$35,897
SWITCH 3COM SERVERS:
$1,282
Structured Cabling:
$17,949

SERVERS HP-UX

SERVERS

$1,047,179

$1,098,461
$3,162,564

SUBNETWORK
$10,068,225
POWER HUB
$24,174,348

POWER HUB
L.A.N. COMMUNICATION
EQUIPMENTS

W.A.N. COMMUNICATION
EQUIPMENT
W.A.N. CABLING
OPTICAL FIBRE

O.S. UNIX

$1,282,051

L.A.N. CABLING TWISTED PAIR

SWITCH 3COM
$6,381,595
Structured Cabling
$12,666,108

Figure 10. Accumulated values in an extract of the asset hierarchy in our case study.
To preserve privacy, the name of the assets and IP addresses have been changed.

In our case study, given that the information has to be constantly available, we
identified two main threats concerning the MAGERIT LAN/WAN Architecture
block of assets (which can be found in Figure 3): T1) physic or logic malfunctions
and T2) main supplies and services interruption. For instance, we found the power
hub in Figure 10 among the assets threatened by T1 and T2. This asset is critical
because most communication systems depend on it and its accumulated value is
high. The countermeasure defined to reduce the risk was C1) Precaution and care
of the communication equipment, which dealt with recommendations for avoiding
malfunctions in communication equipment, such as duplicating strategic elements
and cleaning security areas.
26

2) The selection of the security requirements from the repository related to the
countermeasures defined for each critical asset. The security requirements were
selected according to their priority, their obligement level (compulsory,
recommendable or optional) and, of course, the budget assigned to the development
of the information system. Since the requirements selection was made together with
the stakeholders, we negotiated each requirement at the same time as it was selected.
In the case of the previously mentioned critical asset, the power hub, one of the
security requirements corresponding to the countermeasure C1 which was selected
was:
SyRS.3.5.2.S42. The maintainability contract of the electronic equipment shall include a clause
enforcing the supplier to make a commitment to solve any failure in less than [time in minutes].

If the critical asset had been cheaper than the hub, the security requirement would
have been simply to keep a duplicate of the asset to change it whenever it fails.
The number of requirements from the security profile reused is shown in Table 2. It can
be noted that the percentage of reuse was high, above all in the requirements of the
information layer because it includes the requirements concerning the personal
protection data, which are legally enforced in Spain. Most of the environment
requirements were also selected, since they were already satisfied in the existing
environment (buildings, rooms, furniture, etc.) and therefore their inclusion did not
constitute an additional cost. The requirements of the organizations functions layer were
those least selected because their scope concerns security policies which exceed our case
study.

27

%
Requirements layer

Chosen

Totals

111

136

81.6

Information System

91

121

75.2

Information

48

56

85.7

Organizations functions

29

31

Intangible asset

10

70

266

352

75.6

Environment

TOTAL

Chosen

Table 2 Results of the application of SIREN for security requirements

The use of the SIREN security profile was useful:

in developing a secure information system from the very beginning, taking into
account security issues in the RE process that, otherwise, might be overlooked;

in improving productivity, since there was a decrease in the time dedicated to the
RE process.

The experience has been positive for both parties, since the goal of the research group
was the transference of knowledge from the University to industry in order to validate
academic research in practice, while our regional government counterpart found the
approach useful. They suggested to us the inclusion of a new attribute called responsible
that would indicate the person or team in charge of the realization of the requirement. In
that way the workload of each member of a team could be controlled.
This experience has shown that this approach to the RE process is a good, realistic
starting point for any organization that has not applied any RE method yet. We agree
with Jaaksi [42] in that if the development method is too complicated, the engineers will
not commit themselves to working with it, and a repeatable process cannot be achieved.
A Level 1 organization (according to the REPM, see Section 2.2) would reach easily
Level 2 by applying our SIREN approach.

28

6 Conclusions and further work

This paper presents a practical approach for managing the security of the information
systems from the RE process. We believe it helps in the development of more complete
and robust information systems. This approach is a particularization of a general purpose
process for requirements reuse called SIREN, which shortens the development process
since the analyst starts from a reusable set of requirements. The particularization of
SIREN to the security profile has been based on the risk analysis and management
method MAGERIT (conforming to IEEE 15408 Common Criteria Framework). SIREN
provides a spiral model process based on Kotonya and Sommervilles proposal [10]
which is enriched by explicitly taking reuse into account. The immediate benefit for an
information system development is that the inclusion of these security requirements
ensures that the proper implementation of the system will fulfill the demands of risk
analysis performed with MAGERIT.
Together with the defined process, SIREN provides a requirements specification
documents hierarchy (system, software and tests) based on the IEEE specification
standards. The IEEE 830-1998 standard for SRS has been extended with items from the
VOLERE template [11], but maintaining the conformance with the standard. The
inclusion of a system tests specification document (SyTS), dealing, in particular, with
security, was also found to be of interest, because the SyTS can serve as a checklist
which encompasses all the information system features that must be checked for it to be
considered secure. The traceability relationships between the requirements of the
repository imply inclusive or exclusive dependence relationships. This approach has
been tested in a case study in our regional government and the simplicity of the process
and the fast implication of the staff involved have been proved.

29

As mentioned in Section 2.1, the requirements imposed by the security standards

are generally imprecise. Likewise, we think that some of our security requirements,
expressed in natural language, may also be imprecise. As a first step, we consider that
we can manage imprecision informally through i) the basic, good practice guidelines
[26] (Section 2.2) regarding requirements writing; ii) the Repository Improvement
activity in the SIREN process model (Figure 9); and iii) the fit criteria included in the
SyTS and STS documents (Section 4.1). Nevertheless, we are conscious that complete
imprecision management requires more advanced techniques to formalize both
functional and non-functional requirements. Two lines of future research to tackle this
problem are the following:

In a use case-driven development process, functional requirements will be mapped

into use cases or scenarios. Given that another interest in our research group consists
of the formalization of the UML (Unified Modeling Language) (see [43] and [44]),
further work is required to link the requirements, in particular the most critical, to the
formal models obtained from them, so that they can be rigorously validated and
verified.

Further work is also needed to specify the interactions between security requirements
and other types of non-functional requirements, as well as between functional
requirements. To this extent, we can adopt a formal framework for integrating goals
and goals refinement in requirements models, such as the KAOS methodology
(Knowledge Acquisition in Automated Specification of Software) [45], the GBRAM
methodology (Goal-Based Requirements Analysis Method) [16], or the NFR
methodology (Non-Functional Requirements) [46] in order to capture and evaluate
the alternative goals decomposition. Chung, for example, has already adapted the
NFR framework to the security field [21].

30

Inconsistency management is another interesting aspect to be considered. Inconsistency

can appear, for example, because the domains and profiles are not mutually exclusive.
We think that in well defined profiles such as those coming from MAGERIT and the
personal data protection law, inclusive and exclusive dependencies are enough to avoid
most of the inconsistencies. Nevertheless, inconsistencies cannot be easily avoided in
real applications dealing with multiples and less-formalized perspectives of the system.
Finally, further work will be the link from the SIREN requirements specification
to the rest development artifacts (analysis, design, implementation), since we think that
the integration of SIREN with the rest of the development process will improve the
usefulness of the approach. In this regard, it is necessary to study the combination of
SIREN with other methods, such as the Unified Process [47], and our previous research
dealing with conceptual modeling (the integration of throwaway and evolutionary
prototypes from formal requirements specifications to realize a rigorous and seamless
design [48], and the concurrent definition of use cases and conceptual models from the
business model [49]).
Acknowledgements
This research has been partially financed by the CICYT (Science and Technology Joint Committee),
Spanish Ministry of Science and Technology, project SIRENrm TIC2000-1673-C06-02 SImple REuse of
software requiremeNts and rigorous modeling.
We would like to thank Vicente Rodes from the Regional Ministry of Labor and Welfare in our Regional
Government and Jos Manuel Marqus from the University of Valladolid for their unselfish collaboration
and their useful comments during this research work.

7 References
1. PITAC. (President's Information Technology Advisory Committee). Interim Report to
the President. Setting Federal Research Priorities: Findings and Recommendations.
1998. http://www.ccic.gov/ac/interim

31

2. Lichtestein, S and Swatman, PMC. Effective Internet Acceptable Usage Policy for
Organisations. In. 10th International Conference on Electronic Commerce (Bled'97).
Slovenia. 1997. pp. 503-522.
3. Constitutional Law 15/1999, of December 13, on Protection of private data of
individuals in Spain. (Ley Orgnica 15/1999, de 13 de diciembre, de Proteccin de
Datos de Carcter Personal en Espaa). 1999. (In Spanish)
4. Real Decreto 994/1999, of June 11, in which the Ruling on security measures of
automated files containing private data on individuals is passed. (Real Decreto
994/1999, de 11 de junio, por el que se aprueba el Reglamento de medidas de
seguridad de los ficheros automatizados que contengan datos de carcter personal).
1999: p. 24241. (In Spanish)
5. Infosec. Information Security Breaches Survey. 2000. http://www.infosec.co.uk
6. ISO/IEC Std. 15408. Evaluation Criteria for Information Technology Security. 1999.
7. CCTA, SSADM-CRAMM Subject Guide for SSADM Version 3 and CRAMM Version 2.
Central Computer and Telecommunications Agency, IT Security and Privacy Group,
Her Majesty's Government, London. 1991
8. CLUSIF, MARION version 98. La Commission Mthodes du CLUSIF (Club de la
Securit des Systmes d'Information Franais). 1998
9. MAP.

Metodologa

de

Anlisis

Gestin

de

Riesgos

del

Ministerio

de

Administraciones Pblicas Espaol, MAGERIT v.1.0. 1996. (In Spanish)

10.

Kotonya, G and Sommerville, I, Requirements Engineering. Processes and

Techniques. John Wiley & Sons. 1998

11.

Robertson, S and Robertson, J, Mastering the requirement process. Addison-

Wesley. 1999
12.

Sommerville, I, Software Engineering (6th edition). Pearson Education Limited.

2001
13.

IEEE Std 830-1998. Guide to Software Requirements Specifications (ANSI). The

Institute of Electrical and Electronics Engineers, Inc. IEEE Software Engineering

Standards Collection. 1999.

32

14.

Mili, H, Mili, F, and Mili, A, Reusing Software: Issues and Research Directions.

IEEE Transactions on Software Engineering, 1995. 21(6): p. 528-562.

15.

Cybulsky, J and Reed, K. Requirements Classification and Reuse: Crossing

Domains

Boundaries.

In.

6th

International

Conference

on

Software

Reuse

(ICSR'2000). Springer, Lecture Notes in Computer Science. Viena. 2000. pp. 190210.
16.

Antn, AI and Earp, JB. Strategies for Developing Policies and Requirements for

Secure Electronic Commerce Systems. In. 1st ACM Workshop on Security and Privacy
in E-Commerce (CCS 2000). Athens, Greece. 2000.
17.

Lutz, R. Analyzing software requirements errors in safety-critical embedded

systems. In. Requirements Engineering (RE'93). San Diego, CA. 1993. pp. 126-133.
18.

Skidmore, S, Farmer, R, and Mills, G, SSADM version 4 models and methods,

second edition. NCC Blackwell. 1994

19.

MAP. MTRICA versin 3. Ministerio de Administraciones Pblicas. Secretara

de Estado para la Administracin Pblica. Consejo Superior de Informtica. 2001

(In Spanish)
20.

Policy Framework for Interpreting Risk in eCommerce Security. Technical Report

CERIAS (Center for Education and Research in Information Assurance and Security).
Purdue

University.

1999.

http://www.cerias.purdue.edu/techreports/public/PFIRES.pdf
21.

Chung, L. Dealing with Security Requirements during the development of

Information Systems. In: C. Rolland, F. Bodat, and C. Cauvet (eds.). 5th

International Conference on Advanced Information Systems Engineering (CAISE'93).
Spring Verlag. Paris. Berlin. 1993. pp. 234-251.
22.

Jones, S, Wilikens, M, Morris, P, and Masera, M, Trust Requirements in e-

Business. Communications of the ACM, 2000. 43(12): p. 81-87.

23.

Fenton, N and Neil, M, A Strategy for Improving Safety Related Software

Engineering Standards. IEEE Transactions on Software Engineering, 1998. 24(11):

p. 1002-1014.

33

24.

IEEE/EIA Std 12207.0-1996. Industry Implementation of International Standard

ISO/IEC 12207: 1995. Standard for Information Technology- Software Life Cycle
Processes, in Volume 1: Customer and Terminology Standards. The Institute of
Electrical and Electronics Engineers, Inc. IEEE Software Engineering Standards
Collection. 1999.
25.

ISO 9000 International Standars for Quality Management. 4th edition. ISO

(International Organization for Standarization). 1994

26.

Sommerville, I and Sawyer, P, Requirements Engineering: A good practice guide.

John Wiley & Sons. 1997

27.

Paulk, MC and Weber, CV, The Capability Maturity Model: Guidelines for

Improvement the Software Process. Addison-Wesley. 1995

28.

El Emam, K, Drouin, J-N, and Melo, W, SPICE. The Theory and Practice of

Software Process Improvement and Capability Determination. IEEE Computer

Society. 1997
29.

Baskerville, R, Information Systems Design Methods: Implications for Information

Systems Development. Computing Surveys, 1993. 25(4): p. 375-414.

30.

Gabb, A. The Requirements Spectrum. In. First Regional Symposium of the

Systems Engineering Society of Australia (SE'98). Australia. 1998.

31.

Pressman, RS, Software Engineering. A practitioner's approach. Fifth edition. Mc

Graw Hill. 2000

32.

IEEE/EIA Std 12207.1-1997. Guide for Information Technology - Software life

cycle processes - Life cycle data. The Institute of Electrical and Electronics
Engineers, Inc. IEEE Software Engineering Standards Collection. 1999.
33.

IEEE Std 1233-1998. Guide for Developing System Requirements Specifications.

The Institute of Electrical and Electronics Engineers, Inc. IEEE Software

Engineering Standards Collection. 1999.
34.

Toval, A, Olmos, A, and Rodero, JA. The Role of Requirements Reuse in

Protecting Personal Data. In. 5th World Multiconference on Systemics, Cybernetics

and Informatics (SCI 2001). Orlando, Florida (USA). 2001.

34

35.

Ramesh, B and Jarke, M, Toward reference models for requirements

traceability. IEEE Transactions on Software Engineering, 2001. 27(1): p. 58-93.

36.

Nuseibeh,

B,

Weaving

Together

Requirements

and

Architectures.

IEEE

Computer, 2001. 34(3): p. 115-117.

37.

Vitech Corporation. CORE. 2000. http://www.vtcorp.com

38.

Quality Systems & Software. DOORS. 2000. http://www.qssinc.com

39.

Rational Software. Requisite Pro. 2000. http://www.rational.com

40.

INVESDOC. IECISA. http://www.ieci.es

41.

MAGERIT risk analysis and management in the Regional Information Systems

and Telecommunication Office. Contract CARMMA 3142-UMU. CARM (Autonomous

Community of Murcia) and the Software Engineering Research Group of the
University of Murcia. 1999.
42.

Jaaksi, A, A Method for Your First Object-Oriented Project. Journal of Object

Oriented Programming (JOOP), 1998. 10(8): p. 17-25.

43.

Fernndez, J and Toval, A. Can Intuition Become Rigorous? Foundations for UML

Verification Tools. In: T. FM (eds.). The 11th Int. Sym. on Software Reliability
Engineering. IEEE Computer Press. San Jos (California). 2000. pp. 344-355.
44.

Toval, A and Fernndez, J. Improving System Reliability via Rigorous Software

Modeling: The UML Case. In. 2001 IEEE Aerospace Conference (track 10: Software
and Computing). IEEE Computer Society. Montana, USA. 2001.
45.

Dardenne, A, van Lamsweerde, A, and Fickas, S, Goal-Directed Requirements

Acquisition. Science of Computer Programming, 1993. 20: p. 3-50.

46.

Chung, L, Nixon, B, Yu, E, and Mylopoulos, J, Non-Functional Requirements in

Software Engineering. Kluwer Academic Publishers. 2000

47.

Jacobson, I, Booch, G, and Rumbaugh, J, The Unified Software Development

Process. Addison-Wesley Longman, Inc. 1999

48.

Moros, B, Nicols, J, Garca, J, and Toval, A, Combining Formal Specifications

with Design by Contract. Journal of Object Oriented Programming (JOOP), 2000.

12(9): p. 16-22.

35

49.

Garca, J, Ortn, M, Moros, B, Nicols, J, and Toval, A. Towards Use Case and

Conceptual Models through Business Modeling. In: L.S. Laender AHF, Storey VC
(eds.). 19th International Conference on Conceptual Modeling, ER'2000. Springer.
Salt Lake City (Utah, USA). 2000. pp. 281-294.

36

Figure Legends
Figure 11. A brief summary of our proposal for requirements reuse, particularized to the security field.
The structure of the requirements specification documents is explained in section 4.1.
Figure 2. MAGERIT assets hierarchy.
Figure 3. An extract of the MAGERIT information system layer.
Figure 4. SIREN requirements documents hierarchy.
Figure 5. Reusable requirements repository composed of profiles and domains. Each profile or domain is
a view of the global repository. In the figure, only the SRS document is shown.
Figure 6.. Examples of system (SyRS) and software (SRS) requirements. Labels R1-R4 are only used as
abbreviations.
Figure 7.. Example of exclusive trace between requirements in the SRS. Section 3.4 in the SRS deals
with the design constraints and includes the subsection 3.4.3. Related Applications.
Figure 8.. Spiral model for RE adapted to SIREN.
Figure 9. SIREN approach for requirements reuse.
Figure 10. Accumulated values in an extract of the asset hierarchy in our case study.
To preserve privacy, the name of the assets and IP addresses have been changed.
Table legends
Table 3 Sommerville and Sawyer's top ten guidelines.
Table 2. Results of the application of SIREN for security requirements

37