Electronic Data Interchange System
for Safety Case Management
Alan Eardley
1
, Oleksiy Shelest
2
and Saeed Fararooy
2
1
Staffordshire University, Beaconside, Stafford, ST16 3YL, UK.
2
rcm2 limited, Cheltonian House, Portsmouth Road, Esher, Surrey, KT10 9AA, UK.
Abstract. In this paper the theory of Electronic Data Interchange (EDI) is
applied to the safety case management domain in order to evaluate its benefits
and to assess the potential issues which relate to data exchange within certain
industry sectors. The work is undertaken to establish best practise and to
examine successful techniques identified by other researches in the area of
XML/EDI and to implement them in data transmission from a safety case
management tool based on MS Visio to the Integrated Safety Case
Development Environment (ISCaDE) residing on a Dynamic Object Oriented
Requirements System (DOORS) database. Goal Structuring Notation (GSN)
and its XML dialect Goal Structuring Mark-up Language (GSML) developed at
the University of York is used as messaging specifications to represent the
safety case model. Furthermore, these specifications are used to build EDI
software to transport data across different safety packages.
1 Introduction
The growing popularity of the visual representation of evidence and arguments to
promote safety has led to the development of a range of graphical safety case models,
including
descriptions of the basic elements and relationships in the Goal Structuring Notation
(GSN). The GSN model has been accepted in the safety management field and has
been implemented in a number of computer-based tools (e.g. Wilson et al 1995).
Continuous development work on the model (Kelly 2003) has enabled the first XML
representation of the safety case information Goal Structure Mark-up Language
(GSML).
The aim of this paper is to identify the problems associated with the data interchange
process in safety case management, to develop and evaluate a proposed solution, to
demonstrate any advantage and to examine some potential issues. An XML data
definition for a specific safety case model (GSN) is used to prove data interoperability
between two software environments; Microsoft Visio and Telelogic DOORS.
Kelly (2003) recognizes the relative importance of re-using safety case data and
therefore data interchange in safety case management. Safety cases for complete
Eardley A., Shelest O. and Fararooy S. (2006).
Electronic Data Interchange System for Safety Case Management.
In Proceedings of the 3rd International Workshop on Computer Supported Activity Coordination, pages 116-122
DOI: 10.5220/0002446901160122
Copyright
c
SciTePress
systems are often decomposed into subsystems to manage the complexity of the
models concerned. With this approach it is evident that well-structured safety cases
can be used as templates for future safety arguments in certain application areas. This
feature allows engineers to re-use the repetitive parts of safety arguments. (Kelly
2003).
A number of issues arise in safety case decomposition and potential problems with
separating re-usable patterns and transferring data to other safety cases. It is vital to
consider the interactions between subsystem safety cases, as a disproportionate
amount of effort is devoted to the development of safety cases at a high or ‘overall
level. Thus, problems may arise due to duplication of effort when carrying out safety
tasks and to safety objectives ‘falling through the cracks’ when apportioning safety
responsibilities. (Kelly 2003).
Industry based standards often realise the importance of carrying out safety case
decomposition and understanding the relationship between subsystems. The current
European Railway Standard (CENELEC 1998) describes three relevant types of
safety case:
1 A generic product safety case
that provides evidence that a generic product is safe in a variety of applications;
2 A generic application safety case that provides evidence that a generic product is
safe in a specific class of application;
3 A specific application safety case that is relevant to one specific safety application
only.
The above categorization clearly reflects the need for the successful interchange of
safety case data, which is the theme of this paper.
2 The XML/EDI Approach
With the importance of the exchange of safety case data being established, there is the
opportunity to build a suitable EDI system to facilitate the exchange. The latest
research in the area suggests the use of XML to take advantage of a new generation of
protocols based on the Hypertext Transfer Protocol (HTTP). Such protocols offer
flexibility and improvement to the use of existing ‘legacy’ EDI systems, exchanging
the data in HTTP format via the medium of the Internet, using XML agents (Morrison
2000).
Research by the OASIS group, the UN/EDIFACT initiative, XML/EDI research
group, and CEN Workshop have established that the core elements in building
suitable EDI systems are; a global repository, a business processing logic unit and
messaging specifications (Schmelzer 2002).
The global repository and business processing logic unit are outside the scope of this
paper, but will be discussed briefly in the next section of the paper, which then
focuses on the messaging specification element.
117
3 Technical Implementation
The global repository is briefly discussed with some suggestions made about its
technical implementation. The importance of the business processing logic unit is also
mentioned in this paper. A list of functional requirements for this subsystem was
derived from Tim Kelly research papers and it was suggested to implement the unit in
a form of intelligent agents.
Message syntax is discussed in perspective of XML mapping.
3.1 Global Repository
In practice, global repositories can be databases containing meta-information. A
dedicated repository can contain lists of XML files and their corresponding
descriptions. They can be stored on the Internet or implemented as a web service
accessed through thin and thick clients. In this context Universal Description,
Discovery and Integration (UDDI) technology has potential for improved safety case
data interchange.
Repository maintenance should have a standard organization and processes, although
individual businesses may want to use local copies of a repository to manage their
mission-critical functions (CWA 13993:2000, 2000). In relation to safety
management systems, a repository can be a place to find information about safety
case modelling techniques including Goal Structuring Notation (GSN), Claim
Weighted Factor Analysis (WeFA) and Argument Notation (ASCAD) by Adalard.
Goal Structuring Mark-up Language (GSML) is a representation of GSN safety case
modelling that can use XML dialog data to form XML objects in the proposed
system. The system may contain GSML templates and industry best practices such as
successful safety case arguments represented in GSN.
3.2 Business Logic
The business logic unit should aid the system decomposition process, offer advice on
identifying the subsystem interdependencies and ensure that the system boundaries
are drawn correctly with no unwanted ‘gaps’. Business processes can also be
modelled with various tools (e.g. BP Win and Oracle Process Modeller) and are
applicable to a broad range of services and a common approach is to implement
business logic with the aid of intelligent agents. Such agents can interpret the existing
templates to perform the work needed, interact with the transaction and the user to
create new templates for each new specific task, or identify and obtain alternative
templates for existing jobs. Overall, the business logic should as a minimum take care
of links to system requirements, should ensure there are no unwanted
interdependencies, that there are subsystems and that the boundaries between
subsystems are drawn correctly.
118
3.3 Messaging Architecture
The proposed architecture allows the importation of GSML files into ISCaDE by MS
Visio using GSN with added-on VBA scripts through a DOORS database using the
DOORS Extension Language (DXL). Visio GSN creates a GSML document and
passes it to the ISCaDE GSML ActiveX control. The ActiveX component inherits the
MSXML interfaces and uses them to check if this document is well-formatted and is a
valid XML file. A number of internal operations are performed to map the XML
DOM tree to the GSN model and to derive the resulting GSN objects, attributes and
their relationships. Any GSN-specific information is then passed to the DOORS
database DXL Type library, which communicates directly with the DXL script
through a Windows Component Object Model (COM) interface. The DXL script
stores the data in a DOORS database and creates the relevant modules.
Third party software providers can create industry specific XML subsets and parse
them with ISCaDE GSML objects by querying the ActiveX components for the
supported interfaces. They can incorporate ISCaDE GSML into any software package
that supports the Windows COM architecture or use DXL API to retrieve data from
components and store it in a DOORS database, as in Figure 1
Fig. 1. ISCaDE GSML System Architecture.
4 Case Study
Figure 2 shows an example of a safety case model depicted in MS Visio software.
The safety goal is to prove that fail-safe design will ensure at least 90% of potential
failures that are due to systematic faults are fail-safe. This can be achieved by
breaking the safety case down into sub goals, making relevant assumptions. If sub
goals are proved subsequently the main goal has been proved. Three sub goals are
used:
Flaws in ADC, application software, configuration trip limits and trip logic will
be revealed by dynamic on-line tests;
Compiler, loader and processor flaws are protected against by the reversible
computing technique;
119
Double thermocouple disconnection or veto will cause a trip.
These are used in conjunction with three respective assumptions:
A14 On-line tests detect 90% of systematic failures;
A15 Tests indicate a 99.995% fail-safe bias;
A13 Thermocouples fail low in 90% of cases.
From Figure 2 it can be demonstrated that within the context of the safety
problem and given assumptions about the system, the main goal is proved
Although MS Visio software is effective as a drawing tool it does not depict the
evolution of the safety case within the timeframe, nor does it indicate the resources
used or the changes made to the safety system. However, if the XML structure can be
applied to a safety case, GSML can be used to represent it, as follows:
<goal xlink:type="resource" xlink:label="G.PFD.SYST.2__goal">
<name>G.PFD.SYST.2</name>
<summary>Fail-safe design will ensure
that at least 90% of failures due to
systematic faults are fail-safe
</summary>
</goal>
This is how a goal object will look within this frame:
<solved_by xlink:type="arc" xlink:from="G.PFD.SYST.2__goal"
xlink:to="G.CHECKS__goal" />
This object represents the link:
<assumption xlink:type="resource" xlink:label="A1. __assumption">
<name>A1.</name>
<summary>A15 Tests indicate a
99.995% fail-safe bias
</summary>
</assumption>
Fig. 2. GSN model in Visio.
120
Fig. 3. GSN model in DOORS.
After XML code has been generated, the tool that was described above can transport
the data into the DOORS database. DOORS is a document-based database that allows
its users to trace its objects and view their history or ‘lifecycle’. In other words if
Visio is a graphic approach to safety system specification, then DOORS is a data-
centred approach.
Figure 4 shows how the data was interpreted by ISCaDE that resides on top of
DOORS. As can be seen the communication was complete and effective - no data was
lost or was misinterpreted.
Therefore, by applying this technique companies can benefit from both environments.
Should they require ease of use they can develop their models in MS Visio or should
they be more concerned about traceability features, they can migrate the same data to
ISCaDE or they can use both tools where issues of time or effort are over-ruled by the
demands of safety.
The features of both tools are compared in Figure 5.
Fig. 4. GSN model in ISCaDE.
121
Benefits MS Visio ISCaDE/DOORS
Ease of graphical use Yes No
Enhanced report facilities Yes No
Traceability No Yes
History of events No Yes
Fig. 5. A Comparison of the Features of MS Visio and ISCaDE/Doors.
5 Conclusions
The research highlights the importance of data interchange within the safety-critical
industry and provides sufficient proof of concept to demonstrate that data transfer can
be carried out effectively between existing safety development environments, using
existing EDI systems having the minimum set of components, namely a global
repository, a business logic processing unit and supporting messaging specifications.
The research concentrates on the message processing and building a XML specific
data channel for efficient EDI. GSML is used as a starting point, as its principles can
be expanded to cover all aspects of the safety management industry. These standards
can include ASCAD, the WeFA model, hazard logs and risk analysis matrices. To
ensure future standardisation, it is possible that a regulation body will have to be set
up to validate and authorise proposals for XML representations within EDI structures
in relation to safety cases.
Future work may involve the formation of an independent initiative group across the
industry to research and propose a set of XML-based safety case standards. This will
provide the safety management industry with scientific knowledge that can be used to
develop XML/EDI safety systems.
If approved by the British Standards Committee or ISO (International Organisation
for Standards), these could become industry standards.
References
1. CENELEC (1998) ENV 50129 Railway applications – Safety related electronic systems for
signalling, European Committee for Electrotechnical Standardisation.
2. CWA 13992:2000 (2000) Recommendations for standardisation in the fields of XML for
electronic data interchange, British Standards
3. CWA 13993:2000. (2000) Recommendations and guidance on the use of XML for
electronic data interchange, British Standards.
4. Kelly, T. P. (2003) Managing Complex Safety Cases, Proceedings of the Eleventh Safety-
critical Systems Symposium, Bristol, UK. Springer Verlag.
5. Morrison, M. (2000) XML Unleashed, Sams Publishing
6. Schmelzer, R. (2002) XML and Web Services Unleashed, Sams Publishing Ltd.
7. Wilson, S.P., Kelly T.P. and McDermid, J.A. (1995) Proceedings of the 12
th
Annual CSR
Workshop, York University.
122