engineering process improvement in heterogeneous...
Post on 01-Sep-2018
214 Views
Preview:
TRANSCRIPT
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Institute of Software Technology and Interactive Systems
Engineering Process Improvement in Heterogeneous Multi-Disciplinary Environments
with Defect Causal Analysis
Olga Kovalenko1 Dietmar Winkler1 Marcos Kalinowski2Estefania Serral3 Stefan Biffl1
1TU Vienna, Institute of Software Technology, CDL-Flex, Austria2Federal University of Juiz de Fora, Brazil
3KU Leuven, Belgium
http://cdl.ifs.tuwien.ac.at
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Institute of Software Technology and Interactive Systems
Motivation & Goals
Motivation: Heterogeneous and Multi-Disciplinary Engineering (ME)
Environments. Defects and root causes are hard to find
(even across disciplines)
Key research questions focus on: How to support stakeholders in efficiently
find root causes of defects for future defect prevention? How to implement an improvement strategy with
the defect causal analysis (DCA) approach?
Goals of the paper: Adapted DCA Approach & Evaluation Improvement strategy with DCA
Software Engineer
Process Engineer
Electrical Engineer
Mechanical & process properties
Electrical properties, e.g., Transmission lines.
Software (SW) properties, e.g., Logical Behavior
Risks & Quality Issuesin overlapping areas
1c
1a
1b
2
Pipe &
Instru
mentat
ion
Circuit
Diagram
Functi
on
Block
mech.models
elect. models
SWmodels
2
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Institute of Software Technology and Interactive Systems
Engineering Process Data in ME Projects
Goals in ME Projects:– Consistent and stable engineering data in related disciplines.– Early defect detection and repair.– Defect prevention for future projects based on defect causes.
Traditional and Sequential Engineering Process (derived from our industry partner)
(Manual) Synchronization in Multi-Disciplinary Engineering Environments
Electrical Engineering
Software Engineering
Mechanical Engineering
Dis
cipl
ine-
Spec
ific
Engi
neer
ing
Synchronized Project Data
Defects Repair Sy
nchr
oniz
atio
n
Heterogeneous Data
Identified Defects
!
Consistency CheckingChange Propagation
Defects Detection
3
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Institute of Software Technology and Interactive Systems
Defect Causal Analysis (DCA)
DCA has been successfully applied in Software Engineering*.
Select problem sample
Classify Defects
Identify systematic
errors
Determine main
causes
Develop action
proposals
Document meeting results
1 2 3 4 5 6
*Kalinowski M., Card D.N., Travassos G.H.: “Evidence-based guidelines to defect causal analysis”, IEEE Software 29(4), pp16-18, 2012.
Systematic Process Improvement based on DCA– Expert Workshop involving different stakeholder.– Starting Point: Identified (critical) defects.– Goal: Identifying root causes for defect repair and
prevention.
Defect Classification Schemes Set of attributes that describe a defect, e.g., defined by IEEE, IBM, or HP. Focus on Software Engineering need for extension for multi-disciplinary
and heterogeneous engineering projects.
4
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Institute of Software Technology and Interactive Systems
Research Questions & Solution Approach
Research questions include How to adapt the DCA process to the context of ME projects? How to adapt a defect classification (DC) scheme to the context of ME projects?
Solution Approach Adaptation of the DCA process approach and … … extension of the defect classification scheme … … to address multi-disciplinary engineering projects in heterogeneous environments.
Feasibility Study Initial feasibility study at our industry partner to address most critical root causes based
on DCA findings. Implementation of improvement actions to address root causes. Second study for evaluation of implemented measures.
5
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Institute of Software Technology and Interactive Systems
Adapted DCA Process
Basic Process Steps are similar!
Select problem sample
Classify Defects
Identify systematic
errors
Determine main
causes
Develop action
proposals
Document meeting results
1 2 3 4 5 6
A C B C A B C A C
Communication
Electrical Engineering
Software Engineering
Systematic Error
Mechanical Engineering
Tools
Method/ProcessInputOrganization
People
Adaptation focus on characteristics of multi-disciplinary engineering projects: Different (involved) engineering disciplines (A). Heterogeneous artifacts and data (B). Inter-disciplinary dependencies in project data (C).
Adapted Ishikawa diagram*:
*Ishikawa K.: “Guide to Quality Control”, Asian Productivity Organization Press, 1986.6
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Institute of Software Technology and Interactive Systems
Defect Classification Scheme
Defect Classification (DC) scheme for DCA must cover*: Defect insertion to identify the cause of the defect (1). Defect detection to identify a strategy for defect detection method improvement (2). Defect type (i.e., nature of defect) supporting information for both (1) and (2).
Insertion Context• Discipline• Artifact Type• Artifact• Activity• Project Phase
* Kalinowski M., Card D.N., Travassos G.H.: “Evidence-based guidelines to defect causal analysis”, IEEE Software 29(4), 2012.
Detection Context• Discipline• Artifact Type• Artifact• Activity• Project Phase
Impact
Rating• Priority to fix• Severity (risk)
Current Status within Defect Life Cycle
Defect Type
Defect Mode
Adapted DC scheme based on IEEE consists of 7 attributes:
Different (involved) engineering disciplines (A)Heterogeneous artifacts and data (B)Inter-disciplinary dependencies in project data (C)
7
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Institute of Software Technology and Interactive Systems
Feasibility Study
Study Process Summary Feasibility study in 4 steps to evaluate the adapted DCA process.
Context Definition and Study Planning (1)
DCA Workshops (2) and (4) Involving key stakeholders from our industry partner Focus on the most critical defects.
Lessons Learned: Identification of an improvement strategy (3)
Context Definition
Initial DCA Application
Improvement Implementation
DCA Application(Verification)
1 2 3 4
8
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Institute of Software Technology and Interactive Systems
Context DefinitionFeasibility Study
Context Automation Systems Development Projects,
e.g., Hydro Power Plants. Involvement of various disciplines, e.g., mechanical, electrical,
and software engineering. Isolated tools and data models are not or loosely connected.
9
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Institute of Software Technology and Interactive Systems
Core Defects and Errors Inconsistent Project and Engineering Data. Occur in the End-To-End Test.
Wiring
S2S3
S1
Sensors I/O
Physical Topology PLC Code
VariablesD1D2
Communication
Electrical Engineering
Software Engineering
variable type does not correspond to sensor type
Mechanical Engineering
Tools
Input
Method
Poor testing due to tight time frames
PeopleProgrammer
mistake
Physical topology is out of datePeople
Engineer forgot to notify about changes
Problem in data export
Method
No automated change notification
No automated change propagation
Root Cause Analysis Lack in interoperability of data models and data. No automation supported change propagation/notification
Initial DCA ApplicationFeasibility Study
Proposed Solution (Lessons Learned) Tool-Supported synchronization based on semantic technologies with the ASB*.
*Automation Service Bus (http://cdl.ifs.tuwien.ac.at)10
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Institute of Software Technology and Interactive Systems
Improvement and Second DCA ApplicationFeasibility Study
However … (the results of the second DCA application) Data Model Transformation must be stable and correct. Incorrect transformation rules might lead to (systematic) mapping errors but
they are easier to handle.
Implemented Improvements Synchronization of heterogeneous
data models based on common concepts.
Automated Change propagation / notification consistent project and engineering data.
Systematic Error (identified in a second DCA cycle) Inconsistent engineering data due to incorrect transformations (model transformation
errors)
Improvement Options Additional Quality Assurance Step to verify/validate model transformation.
11
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Institute of Software Technology and Interactive Systems
Summary & Future Work
Summary Multi-Disciplinary Engineering Projects include additional risks due to distributed and
heterogeneous data models that have to be synchronized manually DCA enables the identification of root causes of a certain set of defects systematically. However DCA and Defect Classification approaches, applied in Software Engineering
must be adapted to meet ME projects. A sequence of DCA applications can lead to an improvement strategy applicable in ME
domains.
Future Work Towards tool supported DCA. Large-Scale application and
Case Study in industry environment.
Manual Synchronization
Initial DCA
Inconsistent Project Data
Manual SyncHuman errors(unsystematic)
ASB application
Second DCA
Inconsistent Project Data
Invalid Mapping (Systematic)
Quality Assured ASB
Error
DifferentRoot Causes
QA
Consistent Project Data
Pro
duct
Qua
lity
12
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Institute of Software Technology and Interactive Systems
Thank you ...
Engineering Process Improvement in Heterogeneous Multi-Disciplinary Environments with Defect Causal Analysis
Olga Kovalenko1, Dietmar Winkler1, Marcos Kalinowski2, Estefania Serral3, Stefan Biffl1
1TU Vienna, Institute of Software Technology, CDL-Flex, Austria2Federal University of Juiz de Fora, Brazil
3KU Leuven, Belgium
Dietmar.Winkler@tuwien.ac.at
top related