Visualizing, Analyzing and Optimizing Automotive Architecture Models using Sirius
Advancing digitalization affects almost all aspects of our modern world. A prominent example is that of modern automobiles. From primarily mechanical machines, cars have evolved into driving complex cyber-physical systems over the last decades. Optimizing such systems consisting of vast networks of sensors, actuators, control units, and communication systems is a huge challenge for today's automotive industry and requires standardized and integrated toolchains fit for purpose. Together with a prestigious automotive industry partner, the Technical University of Ilmenau developed an application together with an integrated toolchain for evaluating and optimizing automotive architecture models. This application is based on the Obeo Sirius project as well as the Eclipse Modeling Framework. Based on Sirius, we created a model editor which is used for visualizing, editing, but also analyzing and optimizing automotive models across the boundaries of different architectural layers.
Maximilian Hammer, Technical University of Ilmenau
Maximilian Hammer is a Research Assistant at Technical University of Ilmenau
Visualizing, Analyzing and Optimizing Automotive Architecture Models using Sirius
1. Maximilian Hammer, Ralph Maschotta
and Armin Zimmermann
Systems and Software Engineering Group
Technische Universität Ilmenau
Ilmenau, Germany,
http://www.tu-ilmenau.de/sse/
Model-Driven Application Development for
Evaluation and Optimization of Automotive
E/E-Architectures
SiriusCon 2022
1 Maximilian Hammer, SiriusCon 2022
Visualizing, analyzing and optimizing
automotive architecture models
using Sirius
3. 3
• Formerly: automotive development dominated by mechanical enhancements
• Digitalization new sorts of requirements:
• Safety features
(e.g. driver assistance systems like proximity warning, …)
• Comfort features
(e.g. onboard entertainment systems, rain sensing wipers, …)
• Pro-environmental measures
Introduction
Rationale
Maximilian Hammer, SiriusCon 2022
4. 4
• Modern cars = complex cyber-physical systems (Electric/Electronic systems, short E/E)
• Sensors, Actuators
• Electronic Control Units (ECUs)
• Network switches, bus systems
• … and complexity continuously increases
Demand for suitable evaluation and optimization methods
Introduction
Rationale
Maximilian Hammer, SiriusCon 2022
Source: Vector Informatik GmbH, online
5. 5
• Model-based approaches to handle increasing complexity
• Major challenge: development of flexible, consistent and integrated toolchains
• Research project of Technical University of Ilmenau:
• development of a workflow and integrated toolchain for
• Model-driven analysis, evaluation, optimization of automotive E/E-architectures
Introduction
Rationale
Maximilian Hammer, SiriusCon 2022
6. 6
• This paper presents:
• Central application developed within the project
• Conceptual design and development approach for integrated tools in this context
• Goals:
• Improve reusability and interoperability of applications
• Simplifying integrated and extendable toolchain development
• Toolchains based on uniform metamodels
Introduction
Rationale
Maximilian Hammer, SiriusCon 2022
7. 7
• Central design problem: Deployment Problem
• Find „good“ mapping between logical and physical architecture
• Potentially great impact on efficiency and cost-effectiveness
• Various possible optimization measures like:
• Number of ECUs, overall cable length (cost-effectiveness)
• Mean power consumption (especially important for e-Mobility)
• Communication load balancing
Paper use case: Find optimal communication paths based on existing architecture
Introduction
Scenario
Maximilian Hammer, SiriusCon 2022
8. 8
• Toolchain based on PREEvision by Vector Informatik GmbH
• Model-based automotive E/E engineering
• Proprietary, widely used in the industry
• Presented application based on:
• Eclipse Modeling Framework (EMF)
Open-source modeling environment
• Obeo Sirius Project
Open-source framework for developing graphical model editors
Toolchain & Workflow
Maximilian Hammer, SiriusCon 2022
13. • Next step: context-specific selection
• Determine elements within the context that should be visualized/analyzed
• User defines „anchor function“, i.e. context element
• Is of type LogicalFunction
• Application queries model‘s structure to determine required elements
• Selects LogicalFunctions directly connected to the context function
• Selects ECUs that run these functions
13
Implementation Details
Architecture Visualization
Maximilian Hammer, SiriusCon 2022
15. 15
• Visualization of intermediate elements on all possible communication paths
required
Depth-first search (DFS) algorithm
• Traverses model‘s physical architecture (electronic components and interconnecting bus
systems)
• Stores traversed elements in adjacency lists (also avoids circles)
• If path between source and target is found intermediate elements saved
• After termination: retrieved elements stored in graph-like structure and visualized
Implementation Details
Architecture Visualization
Maximilian Hammer, SiriusCon 2022
18. 18
• Four test models:
• testModel_complete: PREEvision demo model, contains complete E/E architecture
• testModel_compPackages: filtered based on relevant component clusters (e.g. components for
engine control are not relevant for control of windows)
• testModel_context: filtered further based on observation context (i.e. relevant functionalities
within the component cluster)
• referenceModel_useCase: excerpt of real-life automotive model
Performance Evaluation
Maximilian Hammer, SiriusCon 2022
19. 19
• Single bus system = complete graph 𝐾𝐾𝑛𝑛
• 𝐸𝐸𝐾𝐾𝑛𝑛
=
𝑛𝑛∗(𝑛𝑛−1)
2
let 𝑛𝑛 = 20 results in 190 edges
Performance Evaluation
Topology Traversal and Build-Up
Maximilian Hammer, SiriusCon 2022
1436
4
2
44991
1
10
100
1000
10000
100000
testModel_complete
testModel_compPackages
testModel_context
referenceModel_useCase
Figure 1:
execution times
for traversal in ms
Table 1: number of electronic components (EC) and
bus systems (BS) per test model
20. 20
Performance Evaluation
Shortest-Path analysis
Maximilian Hammer, SiriusCon 2022
0,212 0,204 0,195
1,7
0
0,2
0,4
0,6
0,8
1
1,2
1,4
1,6
1,8
testModel_complete
testModel_compPackages
testModel_context
referenceModel_useCase
Figure 2:
execution times for
path analysis in ms
Table 2: number of vertices (V), edges (E) and mapped bus systems
(BSm) contained in the application-internal graph per model
21. 21
• Model-driven analysis, evaluation & optimization of automotive E/E-architectures
• Capabilities:
• Visualization of different architectural layers & mappings
• Editing via graphical model editor
• Architecture analysis & optimization (example: shortest communication paths)
• Easily extendable by introducing certain weighting functions
Potential to enable more sophisticated analyses & optimizations
Conclusion
Maximilian Hammer, SiriusCon 2022
22. 22
• Computational complexity mainly depends on:
• Size of the model
• Complexity of topology (especially bus systems)
Additional bus systems can increase complexity drastically
• Mainly affects Depth-first-search traversal of topology
• Less effect on path analysis due to efficient data preprocessing
Conclusion
Maximilian Hammer, SiriusCon 2022
23. 23
Thank you for your attention!
Maximilian Hammer, SiriusCon 2022
Contact:
Maximilian.Hammer@tu-ilmenau.de
Related paper: Maschotta et al.:
„Model-Driven Aspect-Specific Systems
Engineering in the Automotive Domain“
Systems and Software Engineering Group
Technische Universität Ilmenau,
http://www.tu-ilmenau.de/sse/
Model-Driven Application Development for
Evaluation and Optimization of Automotive
E/E-Architectures