Artemis-2011-1 decision and platform support for model‐based eVolutionary development of Embedded systems Date of preparation

During system engineering, a wide variety of models are created capturing different views of the system. These models are related to one another in various ways. For instance, they can complement one another at the same level of abstraction (e.g. functional view, information view, safety view, and physical view). They can also be in a specification-design relation to one another: One set of models forms the specification and another set of models the design. Often, models can also be simulated and analysed.

This workpackage is about providing a coherent set of methods and tools to manage, visualize, and analyze the wide variety of models used during system development. A working environment for engineers will be the result that allows them to create effectively and efficiently new complex system.

The results of this workpackage are linked to:

Workpackage 3: Here simulation and analysis is done. Workpackage 4 provides methods and tools to maintain the overview of the simulation and to increase the coverage. It also assists in doing the analysis of the models.

Workpackage 5: Here decision support tools are created that help engineers make the proper choices to advance the system under development. These choices are based on the system-models overview provided by workpackage 4.

Workpackage 6: For Longevity, Tuning & Scaling, and Reliability & Safety, it is necessary to have the methods and tools provided by workpackage 4. For the industrial validators, we therefore will fine-tune the working environment to the models needed for development and apply them. For industrial applicability, it then becomes clear how the introduction of these environments will work out for the engineers in the different industries.

T4.1 Viewpoint Management

Lead: MDH

Main contributors: MDH, CRF, ETAS, CAU

Models capturing different viewpoints of systems is key to proper system engineering. This task provides a unifying framework via which the different models used during system engineering can be linked. It is based amongst others on VOSE (Finkelstein et al.’s work on viewpoint oriented systems engineering). The idea is to link functional models (such as Statecharts, CostReduction) to e.g. safety models (using Tropos and Intent) and ensure overall consistency.

Activities:

Identify viewpoints for different industrial validators.

Identify models used for these viewpoints.

Create unifying framework(s).

Develop (visually) supporting methods and tools.

T4.2 Enhanced Visual Modeling

Lead: CAU

Main contributors: ETAS, CAU, CRF

Nowadays, it is common practice that graphical system models are created by “What You See Is What You Get” (WYSIWYG) editors. Even for novices WYSIWYG editors are very easy to use due to their intuitiveness. However, in practice, WYSIWYG editors prove often to be a limiting factor in creating the proper models. The complexity of systems (due to the numbers of components, heavy interaction, etc.) often yields large and unmanageable graphics. In this task we will develop alternative visual editing approaches for fast and effective creation of complex systems models.

Activities:

Identify WYSIWYG models for the industrial validators.

Develop Modeling style guides.

Develop WYSIWYG improvements for modeling and editing such as:

1. 
   In-sync textual and graphical editors based on model-transformation
   
2. 
   Usage of syntactic model structure instead of graphical elements
   
3. 
   Usage of multivariate data tables
   
4. 
   Usage of hierarchical structures
   

Lead: BHL

Main contributors: ETAS, BHL, CAU, RTU

A number of modeling tools provide support for simulation. Usually, you can offer stimuli to the system model and via animation you see the effect of those stimuli. The paradigm that are on offer today do not scale well with large systems. The quantity of interfaces, states and state transitions makes it difficult for the engineer to keep the overview. Furthermore, there are limitations to the exploration of the problem-space resulting often in a limited coverage. This task will focus on creating comprehensible simulation that scales well with the complexity system and that has a proper coverage.

Acitivities:

Identify simulation needs for the industrial validators.

Develop methods and tools to

5. 
   Support the visualization of large amounts of simulation results
   
6. 
   Visualize relevant elements (focus+context)
   
7. 
   Do “Dynamic” model simulation
   
8. 
   Support context-dependent exploration of Data and Dataflow
   
9. 
   Visualize of system behavior and data change over time
   
10. 
    Integrate state-space and event/dataflow exploration
    
11. 
    Enable user to understand relations between parameter settings and results.
    

Lead: BHL

Main contributors: ETAS, NXP-A, AVL, ALM, BHL, RTU, EADS

This task will provide techniques that enable users to understand relations between parameter settings and analysis results. Large state space exploration and cluster exploration for parametric analysis will reduce the need to explore extensively the system simulation results. Moreover, methods as automatic isolation of meaningful analysis results and automatic association of meaning with analysis results will result in a better understanding of system analyses.

Identify analysis needs for the industrial validators.

12. 
    Effectively analyse models.
    
13. 
    Reduce the need for simulation.
    
14. 
    Improve the understanding of the analysis results.
    

• 
  Paul Pop (DTU): I have changed the WP5 title from “Design for Evolvability”, since the WP is not only about design
  
• 
  Paul Pop (DTU)—Question: who will be the WP5 leader? VTT Susanna Teppola – to be decided – Paul will start working on WP description
  
• 
  Paul Pop (DTU)—Question: Will Stig Larsson from MDH continue to be involved in WP5?
  
• 
  Paul Pop (DTU)—Question: What has happened with MDH’s contribution? I have no details! Where and what will they contribute?
  
• 
  From telco minutes: WP5: responsible Paul & Susanna. Objectives are long. Task need descriptions
  
• 
  From telco minutes: The present descriptions are just copies from the Wiki. WP writers should provide new versions before the next telco. This is needed to improve the generic parts of the section descriptions. Descriptions should include the input from the partner questionnaires. Important update would be to change bullet lists into sentences.
  
• 
  From telco minutes: It was asked whether it is possible to describe each WP in a few words. This helps to explain the project better to industrial partners. The descriptions of a few lines in section 3.1 are not easy to understand.
  

The objective of this work package is to develop model-based decision support methods and tools (T5.1), to improve decision-making during product evolutions. The decisions are based on data collection and analysis from previous versions of a product (T5.2). In addition, this work package will propose guidelines and design patterns for building evolvable systems, i.e., systems that are easy to extend (T5.3).

• 
  From telco minutes: Deliverables should be identified. Preferable we have 1-2 deliverables per task. Not more, as we should not overload ourselves with deliverables. One deliverable per task should suffice, but with regard to the two phases, two deliverables may sometimes be needed. Finally, do not give all deliverables the same deadline as they need time to be reviewed and accepted, by the same set of people. A task without deliverable should be integrated with another task.
  

• 
  Paul Pop (DTU): I suggest merging this task with the next one. The task is quite small. The investigation of needs can be done as part of WP1 on requirements and part of the methods and tools task.
  

• 
  Paul Pop (DTU): I have changed the task name, from “Methods for utilizing available knowledge for project and portfolio decision”
  
• 
  Paul Pop (DTU): DTU can be the lead of this task, if nobody else wants to lead it.
  

The state-of-the-art methods in model-based engineering of embedded systems concentrate on the design and implementation, from scratch, of a new system. However, such a situation is uncommon in practice. Typically, a new product is evolved from a previous version. This task will develop model-based decision support methods and tools. When a new product is developed, or a new version is introduced, the decisions taken in the very early stages are critical. The decisions can be on several levels, e.g., business-level, design-level and architecture-level. These decisions will influence the set of possible implementation, with an impact on everything from cost, performance, to energy consumption and reliability. We will improve the quality of decision-making, especially in the early phases.

TEC will work on providing techniques and tool support for handling decision making as well as automatic variability handling and decision resolution. MDH will develop methods for utilizing available knowledge for project and portfolio decision. Contribyte develop, prototype, and test a coherent and flexible tool support for evolutionary design and development of model based products over the life cycle and across related processes and systems. KONECRANES will develop tools for transforming and evolving existing embedded systems to and within the new evolvable model-driven framework. The focus of KONECRANES will be on mobile work machine control, where the developed tools will support the evolving existing embedded mobile work machine control systems, through faster design of embedded control systems and improved re-use of existing systems. RTU and Ldz are focused on the development of a prototype decision support tool, to be applied in the area of intelligent railway transport control, using WP2 models and modelling environment.

Currently, the embedded system architectures are derived without any concern for extensibility. DTU will develop decision support methods and tools for the synthesis of system architectures that are extensible, thus greatly reducing the time and engineering effort required for evolutions. DTU will provide trade-off analysis tools that will allow a systems architect to decide the right amount of extensibility, without compromising other objectives such as performance, cost, energy consumption and dependability. MDH will focus on systems where with parts of the system are being replaced in each version/generation/variant of the product or system. The decision support methods will also include mechanism to separate concerns in complex systems allowing support for evolution (critical partition versus evolvable partitions)

There are different mechanisms that can help ensuring extensibility on software development, e.g., modelling languages and extensions (UML and UML profiles), standardisation initiatives such as Autosar help in gaining extensibility, and validation and verification on models. MU will analyse and develop mechanisms that can be used in MDD in order to obtain extensibility. UoO will develop approaches for supporting a system architect to make appropriate decisions that will result in right amount of extensibility in evolutionary development. DTU will work with WP2 on how to capture the flexibility of a design into the current modelling framework.

T5.2 Data collection, analysis and interpretation for decision support

Notes

• 
  Paul Pop (DTU): I have changed the task name, from “Automation of result analysis and interpretation”
  

When decisions are made (being product management decisions, or architectural and design decisions), the best available information must be used. Decision-making can be improved through better ways to extract, collect, and present information that is requested for well-informed decisions. The focus will be on data from previous versions of the product, previous design iterations, etc., which is currently not used by the existing approaches.

The needs from decision-makers should be made clear in the light of available and emerging modelling techniques, i.e. decision-makers are not always aware of what information actually is available early in the development process. All partners, especially MDH will conduct interviews and surveys to understand what type of information would be needed and useful for decisions throughout development projects. VTT will perform data collection and analysis to understand what type of information would be needed and useful for decisions throughout development projects, to enable effective decision-making in company’s development projects. Often, the data collected presents uncertainties. ISYS will focus their work on providing modelling solution to express such uncertainties and variability. The characterization of uncertainty and variability is an essential input for decision-making support, since it is connected to the risk associated to a decision. Model-based decision-making is a collaborative process. BHL will contribute to the requirements regarding data acquisition in interactive collaboration, and will focus on incorporating, into the tool prototypes developed in this work-package, the information gained from developers’ collaboration.

The extraction of the information must to be automated based on the specified needs from the decision-makers. FHG-HHI will work on post-processing of profiling results, such that the results can be applied and integrated easily in the decision making process. Also, the collection and presentation of the information must be made easily retrievable when required. Almende will focus on automated support for data aggregation and information extraction based on state-of-the-art data mining techniques. This is achieved through a reasoning and learning framework for embedded software systems evolution support, potentially using self-organizing principles. Almende and MDH will work on the derivation of high-level decision support information from low-level simulations with executable models or from run-time generated event traces. The focus of Almende will be on wireless applications for live updates of the analysis and control software for large-scale sensor devices.

To support this data collection, information regarding possible data from WP2, 3 and 4 is needed. Once the data is available, prototypes of how to present data from WP4 can be used for pilots, in workshop formats, and in real development projects (performed in WP6). CG will contribute to the development of the software interface of decision support tools. Their focus will be on intelligent railway transport control system. BHL will work on generating different views (within the model-driven frameworks provided by the rest of the project) that are needed as input for decision-making. NXP-D will evaluate a compact visualization of the regression test results: automated post-procession engine for large data sizes, define report formats for an effective result analysis and interpretation and define decision criteria for result interpretation.

T5.3 Guidelines and patterns for building evolvable systems

Notes

• 
  Paul Pop (DTU): I have changed the task name, from “Evolution of safety runtime frameworks”
  
• 
  Paul Pop (DTU): I have moved here from T5.3 10 PM of NXP-D
  
• 
  Paul Pop (DTU): LDZ’s contribution is quite small, only 2 PM
  

Often, embedded system architectures are derived with little concern for extensibility, rendering evolutions very costly. The methods and tools developed in task 5.1 will support in the creation of systems that are extensible. In this task we will propose guidelines and patterns for building evolvable systems. In addition, we will show (for selected methods and tools proposed in the previous two tasks) how they can be integrated in existing tool flows.

MU will analyse evolution patterns in MDD. Evolution will be considered in the following aspects: functional evolution, QoS evolution, platform evolution, etc. Based on this analysis, MU will propose evolvability guidelines on MDD, i.e., impact of evolution patterns on MDD, mechanisms vs. evolution patterns. MU will provide guidelines for facilitating evolution to meet new quality attribute requirements. This will result in modelling extensions, evolvability guidelines and design patterns. DTU will propose guidelines for developing evolvable safety-critical embedded systems. The focus is on improving the ability to perform upgrades for the non safety-critical components.

The guidelines and patterns proposed in this task are focused on the offline (e.g., before runtime) development of systems. However, evolvability can also be achieved through online (at runtime) adaptation. IKER will deal with evolution of safety grade runtime frameworks for embedded with code generation, models and metamodels.

NXP-D will evaluate the work-package results as part of a new working flow for the design of embedded systems. The plan is that all new product designs and future product platforms will be created making use of the advancements of the new flow. LDZ will evaluate and validate the developed frameworks in the area of railway transport. NSN is developing its process and practices towards evolutionary mode of operation, thus it participates in this task with the purpose to compare its current practices and identify future improvement efforts.

Specification of candidate use-cases as reference models for the validation of DECISIVE techniques in terms of usability of the solution, efficiency, and benefits for the developed products.

Definition of metrics and evaluation techniques

Use cases refinement and implementation running on the DECISIVE reference platform in order to demonstrate the full power of the approach for these cases.

Benchmark of the overall system including methodologies, design tool, and platforms, with the aid of the leading use-cases

Compile the results of the evaluation and provide feedback to technologic-enablers Work Packages

WP6 covers the definition of use cases for the validation of DECISIVE approach. Industrial use-cases will be provided for the validation of DECISIVE approach. Candidate use-cases as reference models for the validation of DECISIVE techniques in terms of usability of the solution, efficiency, and benefits for the developed products will be specified. Metrics and evaluation techniques will be defined. These preliminary use cases will be further refined and implemented running on the DECISIVE reference platform demonstrating the full power of the approach for these cases. Feedback to technology providers Work Packages will be provided. The overall system including the methodologies, design tool, and platforms, will be benchmarked with the aid of these leading use-cases.

Lead: ATEGO

Contributors: CEA?, LDZ , ISYS, PHILIPS

Based on the requirements and initial definition of use case in WP1, we present a detailed design of the implementation of the use cases that are selected for the validation of the DECISIVE approach. For each use case, we will provide a detailed validation plan with rich information that will cover all aspects to evaluate both the functional and non functional behaviour of the solution. The plan will include a detailed sequence of actions to be fulfilled, as well as the hardware and software platforms needed for the evaluation.

In addition, and based on output of the Test Methodology defined on WP1, we also specify a detailed evaluation plan that provides customized metrics and methodology to perform the assessment of the solution for the different domains, while keeping the commonalities within the different use cases. The purpose is to provide common guidelines to drive the evaluation of the different use cases, with a flexible approach that allows to follow a common methodology that can be tailored to the different domains. The metrics will be mainly focused on parameters of the development process rather that the product, although we envisage that the potential benefits of final product will also be assessed.

Lead: PHILIPS

Contributors: EADS, NSN

This task applies DECISIVE methodologies and tools in application fields that deal with longevity products.

Characteristics: High Quality, Product Evolution, Platform, Low-Cost Maintenance

Impact: Faster new products (via evolution) and High utilization through long uptime

Task 6.2.2 Validation for Tuning & Scaling

Lead: CRF

Contributors: ETAS, CG, AVL, NXP-A, NXP-D, CISC

This task applies DECISIVE methodologies and tools in application fields that deal with tuning and scaling of products.

Characteristics: Physical variations in production process , Highly configurable , Calibration needed

Impact: Reduced Fuel Consumption, Low CO₂ emission, Reduced material cost and Enhanced security (of derived products)

Task 6.2.3 Validation for Reliability & Safety

Lead: Atego

Contributors: PAJ, IKERLAN

This task applies DECISIVE methodologies and tools in application fields that deal with guaranteed reliability and safety, involving compliance to safety standards.

Characteristics: Measurement & Control, Safety Critical System, Real-time and Compositional Safety

Impact: Predictable security and safety and Norm Compliance to Safety Standards

Task 6.2.4 Validation for Industrial Applicability

Lead: TDB

Contributors: TBD

This task addresses the deployments of DECISIVE methodologies and tools in industry.

Characteristics: Different way of working , People (and their objections), Incomplete tooling

Impact: Industrial adoption

Task 6.3 Validation of the case-studies

Lead: ISYS

Contributors: ISYS, NXP-D, MU, UES, ???

Once we set up the demonstrator we perform the validation based on the evaluation plan and guidelines specified in Task 6.1 and therefore we may compare the results with the requirements in WP1. The result on this activity will serve as feedback to WP2-WP5, in order to refine DECISIVE methodology and tools.

WP7 has three main objectives which partially determine the success of the DECISIVE project. These are dissemination of the research results, training, exploitation of the project outcomes and standardization activities. More specifically, WP7 will accomplish the following activities:

To create awareness for DECISIVE, and to spread the project outcomes in the industrial and scientific communities.

To continuously disseminate the DECISIVE’s results to technical and scientific audience, as well as to appropriate press and public.

To integrate and package the obtained research results in exploitative form to be used by interested partners, both inside and outside of the consortium.

To aid creation and adoption of DECISIVE methods and tools in industrial product development pilots and environments.

To promote consideration of project results on the respect of standardization activities.

To educate and train next generation researchers…

All the goals should support the ARTEMIS strategic goal to …