Model-based Development for
Vehicular Embedded Systems
Alessio Bucaioni
13-10-2016
STEW 2016
Arcticus Systems
2
OUTLINE
• MESS RESEARCH GROUP
• BACKGROUND
• PROBLEM FORMULATION
• PROPOSED SOLUTION
• UNIQUENESS
• RUNNIN EXAMPLE
• ACCADEMIA-INDUSTRY TRANSFER
3
MODEL-BASED ENGINEERING OF
EMBEDDED SYSTEMS RESEARCH GROUP
16 research projects
15 members
Born in 2011 as a spin-off from the
”Real-Time System Design” group
2 main research areas
4
0
5
10
15
20
25
30
35
2011 2012 2013 2014 2015 2016
Numberofpublications
Years
Conference Paper
Doctoral Thesis
Licentiate Thesis
Book Chapter
Journal Article
MODEL-BASED ENGINEERING OF
EMBEDDED SYSTEMS RESEARCH GROUP
5
Arcticus Systems
MODEL-BASED ENGINEERING OF
EMBEDDED SYSTEMS RESEARCH GROUP
6
BACKGROUND – VEHICULAR EMBEDDED
SYSTEMS
PARENTAL
CONTROL
WINDSHIELD
WIPER
CONTROL
ENGINE
CONTROL
AIRBAG
DEPLOYMENT
ADAPTIVEFRONT
LIGHTING
ADAPTIVECRUISE
CONTROL
AUTOMATIC
BRAKING
ELECTRICPOWERSTEERING
ELECTRONIC
THROTTLE
CONTROL
ELECTRONICVALVE TIMING
IDLE STOP/START
CYLINDER
DE-ACTIVATION
ACTIVE
VIBRATION
CONTROL
OBDII
REMOTE
KEYLESS
ENTRY
BLINDSPOT
DETECTION
LANE
DEPARTURE
WARNING
TRANSMISSIONCONTROL
SEATPOSITION
CONTROL
ACTIVEYAW
CONTROL
PARKING
SYSTEM
ELECTRONIC
STABILITY
CONTROL
ANTILOCK
BREAKING
TIREPRESSURE
MONITORING
NIGHT
VISION
HEAD-UP
DISPLAY
DRIVERALERTNESS
MONITORING
INSTRUMENT
CLUSTER
ACCIDENT
RECORDER
EVENTDATA
RECORDER
AUTO-DIMMING
MIRROR
INTERIOR
LIGHTING
ACTIVECABINNOISE
SUPPRESSION
VOICE/DATA
COMMUNICATION
CABINENVIRONMENT
CONTROLS
DSRC
ENTERTAINMENTSYSTEMS
BATTERYMANAGEMENT
LANECORRECTION
ELECTRONIC
TOLLCORRECTION
DIGITALTURN
SIGNALS
NAVIGATIONSYSTEM
SECURITYSYSTEM
ACTIVEEXHAUST
NOISESUPPRESION
RIGENERATIVE
BREAKING
ACTIVESUSPENSION
HILLHOLD
CONTROL
Courtesy of www.volvo.com
7
BACKGROUND - VEHICULAR
EMBEDDED SYSTEMS
“More than 80 percent of
vehicle innovation comes
from embedded systems”
- MANFRED BROY
Professor of informatics at Technical University, Munich
8
0
50000000
100000000
150000000
200000000
250000000
300000000
350000000
Late 1970s Nowadays
Linesofcodes
Years
Size of vehicular embedded software
BACKGROUND - VEHICULAR
EMBEDDED SYSTEMS
9
BACKGROUND - VEHICULAR
EMBEDDED SYSTEMS
Courtesy of www.bmw.com
4,5 times more expensive
Multi-core platforms
25% longer schedules
3 times as many software engineers
6
* S. Balacco, C.Rommel. Next Generation Embedded Hardware Architectures:Driving Onset of Project
Delays, Costs Overruns and Software Development Challenges. Klockwork Inc. 2010.
BACKGROUND - VEHICULAR
EMBEDDED SYSTEMS ON MULTICORE
BACKGROUND - MODEL-DRIVEN
ENGINEERING
11
- BRAN SELIC
Father of Real-Time UML
“As our systems grow in
complexity traditional code-
centric development methods
are becoming intractable”
BACKGROUND - MODEL-DRIVEN
ENGINEERING
12
Abstraction
Automation
+
=
Model-driven Engineering
BACKGROUND – EAST-ADL
13
14
Vehicle Level
Analysis Level
Design Level
Implementation
Level
Activities Abstraction levels Format
Capture requirements on
E2E vehicle functionality
Consistency analysis of requirements.
Functional verification
Prototyping, system properties,
timing and resource analysis.
Complete SW architecture
Modelling of features.
SW architecture, HW architecture, SW
to HW allocation,
Often informal. Textual.
Solution-independent
Formal, model-based.
Allocation independent
Formal, model-based.
Implementation-independent.
Formal, model-based.
Implementation details.
BACKGROUND – EAST-ADL
15
PROBLEM FPRMULATION
- PONTUS DE LAVAL
CTO at Saab AB
“It is so much cheaper to find
defects at design time”
16
PROPOSED SOLUTION - MY
RESEARCH IN A NUTSHELL
Model-based software development
methodology which supports early timing
analysis for vehicular embedded systems.
Design Level
Implementation
Level
Timing analysis
17
PROPOSED SOLUTION -
METHODOLOGY
Analysis
results
M2M
transformation
Timing analysis
& filter
Analysis
results
M2M in-place
transformation
DesignlevelImplementationlevel
EAST-ADL
design model
u-Rubus
model
u-Rubus model
with
analysis results
Negative
feedback
18
UNIQUENESSES – WHAT DO YOU
GAIN ?
• Reduce accidental complexity
• Early timing verification
• Support uncertainty
• Support for multi-core
RUNNING EXAMPLE: INTELLIGENT PARKING
ASSIST
19
Proximity_Sensor_DFP Input_Process_DFP Path_Calculator_DFP CAN_Send_DFP CAN_Receive_DFP Control_DFP Brake_Actuator_DFP
IPAssistant_DFP Actuator_DFP
15 ms
20 ms
20
(1)
(2)
(3)
(4)
Software Circuit Clock
Connector data
Connector trigger
Data ports
Trigger ports
Timing constraints
Timing constraints
RUNNING EXAMPLE: INTELLIGENT PARKING
ASSIST
Reduce complexity
Support uncertainty
21
Timing analysis has filtered the solution space.
However there are still 14 RCM models to inspect.
(1)
(2)
(3)
Software Circuit Clock Connector trigger Trigger ports
RUNNING EXAMPLE: INTELLIGENT PARKING
ASSIST
Early timing
verification
22
Support uncertainty
RUNNING EXAMPLE: INTELLIGENT PARKING
ASSIST
23
METHODOLOGY FOR MULTICORE
START
Functional Model
RubusMM_SW
Platform Model
RubusMM_HW
M2M Transformation
JTL
Execution Models
μ-RubusMM_SW + Timing
Model-based Timing
Analysis
Modify the
Allocation Models
Modify the
Functional Model
Code Generation
END
Are the Timing
Requirements Met?
Is It a Single-core
Platform?
Are all the Allocations
Model checked?
YES
NO
NO
YES
YES
NO
24
MEES CONTRIBUTIONS
Vehicle Level
Analysis Level
Design Level
Implementation
Level
Abstraction levels Contribution of the MEES research group
finished contribution ongoing contribution
Rubus Component Model (RCM)
RCM metamodel definition (RubusMM)
Exact RTA
RTA for CAN and high level protocol, e.g., HCAN, CANopen
E2E response time
E2E delays, e.g., age and delay
Shared stack analysis
Switched ethernet
SWEET benchmark
Extensions for multi-core platforms
RubusMM extensions for multi-core platforms
Model-based methodology for early predictability
Predictability enabled on design assumptions
Predictability enabled for legacy nodes
RubusEASTandtranslationofTADL2constraints
25
ACADEMIA-INDUSTRY TRANSFER
MDH
BASEMENT
SaveComp
ProSave
EMDEF
FEMMVA
SynthSoft
RCM 1&2
RCM 3
RCM 4
Extension of timing
analysis, modelling
support
Multicore
Arcticus
1994
1996
2002
2005
2005
2012
2009
2012
2014
2014
2018
RCM 4 +
26
ACADEMIA-INDUSTRY TRANSFER
Arcticus Systems
Requirements,
Existing tools,
Certified RTOS
Methods, Technbiques,
Prototypes
Thank you for the attention!
Questions?

Model-based Development for Vehicular Embedded Systems

  • 1.
    Model-based Development for VehicularEmbedded Systems Alessio Bucaioni 13-10-2016 STEW 2016 Arcticus Systems
  • 2.
    2 OUTLINE • MESS RESEARCHGROUP • BACKGROUND • PROBLEM FORMULATION • PROPOSED SOLUTION • UNIQUENESS • RUNNIN EXAMPLE • ACCADEMIA-INDUSTRY TRANSFER
  • 3.
    3 MODEL-BASED ENGINEERING OF EMBEDDEDSYSTEMS RESEARCH GROUP 16 research projects 15 members Born in 2011 as a spin-off from the ”Real-Time System Design” group 2 main research areas
  • 4.
    4 0 5 10 15 20 25 30 35 2011 2012 20132014 2015 2016 Numberofpublications Years Conference Paper Doctoral Thesis Licentiate Thesis Book Chapter Journal Article MODEL-BASED ENGINEERING OF EMBEDDED SYSTEMS RESEARCH GROUP
  • 5.
    5 Arcticus Systems MODEL-BASED ENGINEERINGOF EMBEDDED SYSTEMS RESEARCH GROUP
  • 6.
    6 BACKGROUND – VEHICULAREMBEDDED SYSTEMS PARENTAL CONTROL WINDSHIELD WIPER CONTROL ENGINE CONTROL AIRBAG DEPLOYMENT ADAPTIVEFRONT LIGHTING ADAPTIVECRUISE CONTROL AUTOMATIC BRAKING ELECTRICPOWERSTEERING ELECTRONIC THROTTLE CONTROL ELECTRONICVALVE TIMING IDLE STOP/START CYLINDER DE-ACTIVATION ACTIVE VIBRATION CONTROL OBDII REMOTE KEYLESS ENTRY BLINDSPOT DETECTION LANE DEPARTURE WARNING TRANSMISSIONCONTROL SEATPOSITION CONTROL ACTIVEYAW CONTROL PARKING SYSTEM ELECTRONIC STABILITY CONTROL ANTILOCK BREAKING TIREPRESSURE MONITORING NIGHT VISION HEAD-UP DISPLAY DRIVERALERTNESS MONITORING INSTRUMENT CLUSTER ACCIDENT RECORDER EVENTDATA RECORDER AUTO-DIMMING MIRROR INTERIOR LIGHTING ACTIVECABINNOISE SUPPRESSION VOICE/DATA COMMUNICATION CABINENVIRONMENT CONTROLS DSRC ENTERTAINMENTSYSTEMS BATTERYMANAGEMENT LANECORRECTION ELECTRONIC TOLLCORRECTION DIGITALTURN SIGNALS NAVIGATIONSYSTEM SECURITYSYSTEM ACTIVEEXHAUST NOISESUPPRESION RIGENERATIVE BREAKING ACTIVESUSPENSION HILLHOLD CONTROL Courtesy of www.volvo.com
  • 7.
    7 BACKGROUND - VEHICULAR EMBEDDEDSYSTEMS “More than 80 percent of vehicle innovation comes from embedded systems” - MANFRED BROY Professor of informatics at Technical University, Munich
  • 8.
  • 9.
    9 BACKGROUND - VEHICULAR EMBEDDEDSYSTEMS Courtesy of www.bmw.com
  • 10.
    4,5 times moreexpensive Multi-core platforms 25% longer schedules 3 times as many software engineers 6 * S. Balacco, C.Rommel. Next Generation Embedded Hardware Architectures:Driving Onset of Project Delays, Costs Overruns and Software Development Challenges. Klockwork Inc. 2010. BACKGROUND - VEHICULAR EMBEDDED SYSTEMS ON MULTICORE
  • 11.
    BACKGROUND - MODEL-DRIVEN ENGINEERING 11 -BRAN SELIC Father of Real-Time UML “As our systems grow in complexity traditional code- centric development methods are becoming intractable”
  • 12.
  • 13.
  • 14.
    14 Vehicle Level Analysis Level DesignLevel Implementation Level Activities Abstraction levels Format Capture requirements on E2E vehicle functionality Consistency analysis of requirements. Functional verification Prototyping, system properties, timing and resource analysis. Complete SW architecture Modelling of features. SW architecture, HW architecture, SW to HW allocation, Often informal. Textual. Solution-independent Formal, model-based. Allocation independent Formal, model-based. Implementation-independent. Formal, model-based. Implementation details. BACKGROUND – EAST-ADL
  • 15.
    15 PROBLEM FPRMULATION - PONTUSDE LAVAL CTO at Saab AB “It is so much cheaper to find defects at design time”
  • 16.
    16 PROPOSED SOLUTION -MY RESEARCH IN A NUTSHELL Model-based software development methodology which supports early timing analysis for vehicular embedded systems. Design Level Implementation Level Timing analysis
  • 17.
    17 PROPOSED SOLUTION - METHODOLOGY Analysis results M2M transformation Timinganalysis & filter Analysis results M2M in-place transformation DesignlevelImplementationlevel EAST-ADL design model u-Rubus model u-Rubus model with analysis results Negative feedback
  • 18.
    18 UNIQUENESSES – WHATDO YOU GAIN ? • Reduce accidental complexity • Early timing verification • Support uncertainty • Support for multi-core
  • 19.
    RUNNING EXAMPLE: INTELLIGENTPARKING ASSIST 19 Proximity_Sensor_DFP Input_Process_DFP Path_Calculator_DFP CAN_Send_DFP CAN_Receive_DFP Control_DFP Brake_Actuator_DFP IPAssistant_DFP Actuator_DFP 15 ms 20 ms
  • 20.
    20 (1) (2) (3) (4) Software Circuit Clock Connectordata Connector trigger Data ports Trigger ports Timing constraints Timing constraints RUNNING EXAMPLE: INTELLIGENT PARKING ASSIST Reduce complexity Support uncertainty
  • 21.
    21 Timing analysis hasfiltered the solution space. However there are still 14 RCM models to inspect. (1) (2) (3) Software Circuit Clock Connector trigger Trigger ports RUNNING EXAMPLE: INTELLIGENT PARKING ASSIST Early timing verification
  • 22.
    22 Support uncertainty RUNNING EXAMPLE:INTELLIGENT PARKING ASSIST
  • 23.
    23 METHODOLOGY FOR MULTICORE START FunctionalModel RubusMM_SW Platform Model RubusMM_HW M2M Transformation JTL Execution Models μ-RubusMM_SW + Timing Model-based Timing Analysis Modify the Allocation Models Modify the Functional Model Code Generation END Are the Timing Requirements Met? Is It a Single-core Platform? Are all the Allocations Model checked? YES NO NO YES YES NO
  • 24.
    24 MEES CONTRIBUTIONS Vehicle Level AnalysisLevel Design Level Implementation Level Abstraction levels Contribution of the MEES research group finished contribution ongoing contribution Rubus Component Model (RCM) RCM metamodel definition (RubusMM) Exact RTA RTA for CAN and high level protocol, e.g., HCAN, CANopen E2E response time E2E delays, e.g., age and delay Shared stack analysis Switched ethernet SWEET benchmark Extensions for multi-core platforms RubusMM extensions for multi-core platforms Model-based methodology for early predictability Predictability enabled on design assumptions Predictability enabled for legacy nodes RubusEASTandtranslationofTADL2constraints
  • 25.
    25 ACADEMIA-INDUSTRY TRANSFER MDH BASEMENT SaveComp ProSave EMDEF FEMMVA SynthSoft RCM 1&2 RCM3 RCM 4 Extension of timing analysis, modelling support Multicore Arcticus 1994 1996 2002 2005 2005 2012 2009 2012 2014 2014 2018 RCM 4 +
  • 26.
    26 ACADEMIA-INDUSTRY TRANSFER Arcticus Systems Requirements, Existingtools, Certified RTOS Methods, Technbiques, Prototypes
  • 27.
    Thank you forthe attention! Questions?