07.05.

2024
Thomas Hülshorst, Akshay Bujone, Christian Kirchhof,
Marius Wegener, Jan Hake
Navigating the Shift to
Prepared for Centralized E/E Architectures in SDVs
MATHWORKS AUTOMOTIVE
CONFERENCE

© by FEV – all rights reserved. Confidential – no passing on to third parties

The industry is transforming towards
software-centric products

• Advanced Features
Market
• Updateability
Demand
Seamless smartphone • Upgradeability
integration
Connected High-performance
entertainment
Advanced • Simplified, modular hardware
ADAS features Cost
• Cost efficient SW platforms
Reduction

Electric
• Development efficiency by
New features
DevOps, automation,
over lifetime Continuous
virtualization
improvement
Vehicle always of the vehicle
“up-to-date” Competition • Reduced time-to-market and;
new way of working
• Reduced recalls (by OTA and
data analytics)

2
© by FEV.io – all rights reserved. Confidential – no passing on to third parties
Software-Defined Vehicles (SDVs)
VALUE FOR CUSTOMERS IS CLEAR… …BUT TECHNICAL BACKBONE IS NOT READY
“When I bought a car in the past, the day I got it was the Traditional E/E & SW Architecture:
best the car would ever be. With a software-defined car, Decentralized & signal-oriented (monolithic)
the day you buy it is the worst it will ever be.
From the point forward, however, it’s going to be like magic.” ECU & SW
ECU & SW ECU & SW

ECU & SW ECU & SW

ECU & SW ECU & SW

Jensen Huang
ECU & SW
ECU & SW ECU & SW
ECU & SW

CEO NVIDIA ECU & SW

ECU & SW
ECU & SW
ECU & SW

ECU & SW ECU & SW

ECU & SW

Seamless smartphone
integration VISION
Connected High-performance (IN-VEHICLE)
entertainment
Advanced Centralized E/E Architecture Mainly service-oriented
ADAS features w/ high-performance software architecture
computing cluster
SW SW SW SW SW
High-speed Feature Feature Feature Feature Feature
Zonal connection Zonal
HPC HPC SW SW SW SW SW
Service Service Service Service Service

New features Central SW SW SW SW SW

over lifetime Continuous HPC Service Service Service Service Service

OS, Hypervisor, Containers

improvement Zonal Zonal & Middleware
Vehicle always of the vehicle HPC High-speed HPC
“up-to-date” connection High-Performance Computing

3
© by FEV.io – all rights reserved. Confidential – no passing on to third parties
Our legacy: FEV has a wide range of cross-domain control features and well
established automotive-grade processes, methods and toolchains

Battery
FEV AUTOMOTIVE CONTROL SOFTWARE LIBRARY Managem
System
2WD 4WD
1-2 2-4
Mot Mot

P0
Motor
Electric Control
Vehicle

Automotive Software
Unit
Powertrain Control

P1

FEV Core Predictive

Parallel
P2
Hybrid Control Energy
Managm.
Software

P3

Mixed
ADAS/AD
Hybrid

P4

Power
DHT PxPy E-PGS Intelligent
Split
Chassis
Control

4
© by FEV.io – all rights reserved. Confidential – no passing on to third parties
Our legacy: FEV has a wide range of cross-domain control features and well
established automotive-grade processes, methods and toolchains

Battery
FEV AUTOMOTIVE CONTROL SOFTWARE LIBRARY Managem
Feature Examples: System Feature Examples:
2WD 4WD
1-2 2-4
Mot Mot State of Charge /
Thermal
State of Health
Management
P0
Motor
Electric Control
Vehicle

Automotive Software
Unit
Field-oriented
Powertrain Control

Energy & Charge

P1 Control
Management

FEV Core Predictive

Parallel
Torque P2
Hybrid Control Energy Range Prediction
Managm.
Management Software

P3 Adaptive Cruise
Vehicle
Control
Coordination Mixed
ADAS/AD
Hybrid

P4
Traffic Jam
Power
Diagnostics DHT
Split
PxPy E-PGS Intelligent Chauffeur
Chassis
Control

Powered-by 5
© by FEV.io – all rights reserved. Confidential – no passing on to third parties
What about the relocation of features/functions?
Major impact on function development
DEVELOPMENT FROM EMBEDDED CONTROLLER TO VIRTUALIZED PLATFORMS

Conventional E/E System Software-defined E/E System

Functions hosted at dedicated Functions independent from dedicated controllers,

embedded controllers hosted on Cloud/Central/Zonal or Embedded target

Signal-oriented Verification/Validation mainly Software organized in

Software architectures Virtualized platforms
at embedded controller level Service-oriented demand strong shift-left
architectures (SOA) and virtual verification and
validation
SOA = Service Oriented Architecture, SOP = Start of Production
Source: FEV 2
© by FEV.io – all rights reserved. Confidential – no passing on to third parties
Feature reallocation: What is the benefit?
FROM TRADITIONAL ECU TO THE CLOUD

CLOUD ENVIRONMENT

• PERFORMANCE: UNLIMITED
HPC
ON-BOARD HPC − Performance can be used
whenever requested

• PERFORMANCE: IMPROVED • DATA-ACCESS: HIGH

ON-BOARD ECU − Computational power can be − Access to large sets of data
allocated to algorithm when − Data-driven models can be
needed adapted fleet wide
• PERFORMANCE: LIMITED
− ECU is not suited for highly • DATA-ACCESS: MEDIUM • AVAILABILITY: MEDIUM
complex or data-driven − Access to almost all on-board − Cloud is widely available
algorithms data (navigation, HVAC,
powertrain) is available − Resources can be scaled
• DATA-ACCESS: LOW whenever they are needed
• AVAILABILITY: HIGH based on actual customer
− ECU is limited to data received request
via bus communication − HPC is permanently available
− In case of in-availability the
− Can be used as fallback solution local HPC compute can take
• AVAILABILITY: HIGH
when cloud is not available over
− ECU is always available for
calculation

7
© by FEV.io – all rights reserved. Confidential – no passing on to third parties
Feature reallocation: What is important to remain automotive-grade?
FROM TRADITIONAL ECU TO THE CLOUD

ON-BOARD ECU

• Hard Real-Time: HIGH HPC

ON-BOARD HPC
− Hard Real-Time requirements
are Standard and can be met
with existing solutions • Hard Real-Time: MEDIUM
− Cross-domain applications with
CLOUD ENVIRONMENT
• Functional Safety: HIGH
mixed criticality regarding hard
− Functional Safety requirements real-time
are standard and can be met
• Hard Real-Time: LOW
with existing solutions • Functional Safety: MEDIUM − Not automotive standard
− Cross-domain applications with
mixed criticality regarding
• Functional Safety: LOW
Functional safety − Not automotive standard
− Up to ASIL-D possible, but higher − Cloud infrastructure not
effort available for ASIL-B/C/D today

8
© by FEV.io – all rights reserved. Confidential – no passing on to third parties
Which functions are candidates for
re-allocation?

Vehicle Motion, e.g. electric motor control Predictive Energy Management, e.g.

Field-oriented Control Range Estimation

• Strict real-time requirements • No strict real-time requirements

• Functional safety relevant • No functional safety requirements

• Implementation is independent of
• Implementation required to be
controller hardware
adapted to controller hardware
• Functionality can be improved and
• Fixed functionality that is not
updated in the field
expected to change Keep it on local
ECU! • Strong benefit of access to large sets
of data
Reallocation • Data-driven models can be adapted
recommended! fleet wide

© by FEV.io – all rights reserved. Confidential – no passing on to third parties

CUBE –The 360° Philosophy of FEV represents a mindset and a Systems
Engineering process model to realize complex SDV enabled systems
CUBE UNITES ALL DISCIPLINES INTO A STRUCTURED APPROACH TO TACKLE HOLISTIC PRODUCT DEVELOPMENT

CUBE
• Unites all development disciplines under
Eco System one common approach to realize complex,
Decomposition

Product (Vehicle)
software driven products
Domain

Sub Domain SW
EE
SW
EE
• Creates a comprehensive perspective on
Mech Mech
Component
products, with clear technical
System Discipline Discipline System responsibilities
Specific
Design Specific Integration
Industrialization
Design
• Focuses on collaboration to manage the
Abstraction
work efficiently

CONSISTENT SYSTEMS ENGINEERING

BENEFIT FOCUS ON NON-FUNCTIONAL REQ. FOR SDV DECISION

FOR SDV EFFICIENT MANAGEMENT OF LEGACY & OTS ELEMENTS

SEAMLESS HANDOVER TO SW ARCHITECTURE DESIGN

© by FEV.io – all rights reserved. Confidential – no passing on to third parties

 High level Service Oriented Architecture definition methodology
SYSTEM DESIGN PRODUCT SPECIFIC DESIGN

A: FEATURE ANALYSIS B: SYSTEM ARCHITECTURE C: TOPOLOGY MAPPING D: N/W BINDING E: N/W DESIGN

REQUIREMENTS FUNCTION BEHAVIOR DECOMPOSITION PROTOCOLS EEA

1 5 9 13
DOMAIN Reusability & Sub-
Service or
ANALYSIS System Dependency Client/Server or
Signal / S2s Decision
Analysis Sender/Receiver

2
6 10
Real Time or High Methods, Events, Fields
PERFORMANCE performance or Signals
14
SOME/IP, DDS,
MACsec, VLAN,
AVB/TSN, HTTP/REST,
3 MQTT
7 11 and CAN, LIN
ASIL and Secure Assent
FUSA / CYSEC E2E Profile, SecOC
Data Definition

15 Service ID, Instance ID,

Method / Field Event ID
4 8 12 CAN-ID, LIN ID
NETWORK / HW Message-Signal Layout
Network Topology Datatypes, Periodicity
CONSIDERATIONS

| 11
© by FEV.io – all rights reserved. Confidential – no passing on to third parties
A high-performing E/E architecture builds the backbone for future product
and development excellence – different levels of centralization are possible
E/E ARCHITECTURE STRATEGY • Illustrative

SDV CORE (“VEHICLE BRAIN”)

APPLICATIONS

OEM API

MIDDLEWARE

Cloud
OS/ VIRTUALIZATION

HIGH-PERFORMANCE COMPUTER

Option 1 Option 2 Option 3

ECU ECU ECU ECU ECU ECU
Domain-Centralized Controls Central + Zonal HPCs Central HPC
Connectivity Zonal
High-speed
Zonal
ECU ECU ECU ECU ECU ECU
CENTRALIZATION
connection

HPC Controller Controller Vehicle Computer

APPROACH
ADAS/AD Infotainment Vehicle or
ECU ECU ECU ECU ECU ECU
HPC HPC Computer
Vehicle Vehicle
ECU ECU ECU Chassis Powertrain Zonal Zonal Computer 1 Computer 2 ECU ECU ECU
ECU High-speed ECU Controller High-speed Controller
connection connection

Top-level control
through “Vehicle brain”
Traditional EEA SDV Traditional EEA
High performance and
classical controllers
Consideration of legacy through
Zonal controllers to
stepwise evolution towards desired centralization level minimize wiring

Note: Vehicle Computer, Zonal Controller = System on a Chip

12
© by FEV.io – all rights reserved. Confidential – no passing on to third parties
What is the right architecture for my use-cases?
Software architecture classification (simplified for illustration)
OTA Legacy Software
Updates Migration Reuse
APP APP APP APP APP APP

VIRTUAL MACHINE ARCHITECTURE

COMBINATION OF ARCHITECTURES
I • Execute virtual machines on HPC BASE
SOFTWARE
Drivers Drivers

• Emulate legacy software “as is” OS OS

EMULATOR

• Best for legacy migration and safety VIRTUAL

MACHINE
VIRTUAL
MACHINE
VIRTUAL
MACHINE

CONTAINERIZED ARCHITECTURE

II • Facilitate over-the-air updates better APP APP APP APP APP APP APP APP

• Supports up-date of single features CONTAINER ENGINE

• Leaner admin approach

3RD 3RD 3RD 3RD
OEM OEM OEM OEM
PARTY PARTY PARTY PARTY
SERVICE-ORIENTED ARCHITECTURE APP APP APP APP APP APP
APP APP

OEM OEM OEM OEM

• Approach maximize Software re-use
III
OEM API
APP APP APP APP

• Supports best 3rd party app ecosystem Core Core Core Core Core
Service #1 Service #2 Service #3 Service #4 Serv

MIDDLEWARE

13
© by FEV.io – all rights reserved. Confidential – no passing on to third parties
FEV SDV - Core platform - Current demo setup based
on SDV Reference Architecture Type Existing ADAS and Powertrain
control features ported via VM

CORE PLATFORM IVI CONTROLLER

Flexible
communication VM-VCU VM ADAS VM-Digital VM-
Cluster Infotainment
DISPLAY VCU ADAS
CLUSTER IVI
TRQ ACC
Virtual Bus (Some.IP / DDS)
Adaptive Adaptive Fixed 3rd
Autosar Autosar
Party OS
QNX OS AGL AGL ANDROID

QNX Hypervisor (customized)

HDMI USB
R-CAR H3 HPC SOC i.MX8
USB HDMI

VEHICLE SIMULATION
JOYSTICK
Testing in Ethernet
Simulation Switch DISPLAY
Internet in
AA: Autosar Adaptive; IVI: In vehicle Infotainment
© by FEV.io – all rights reserved. Confidential – no passing on to third parties
Overview global FEV SDV initiative for a software-driven product
development including a collaborative engineering framework

FEV SDV CORE PLATFORM BUILD UP FEATURE DEVELOPMENT BASED ON FEV IP

• Cross-domain applications
• HPC (Renesas, NXP,
Qualcomm)
• Hypervisor (QNX, GHS, ..)
• FEV Virtual Bus ADAS/AD INFOTAIN- VEHICLE ENERGY /
BODY
• OTA Enabled MENT MOTION BATTERY
• Vehicle API CONTROL
• Classic & Adaptive enabled
FEV SDV platform demo / CES 2024

DEVELOPMENT WORKBENCH SDV CLOUD & DATA CENTER BUILD UP

COLLABORATION PROCESSES SW MGMT CI/CD

VIRTUAL ENGINEERING FRAMEWORK DIGITAL TWIN DATA MGMT. CLOUD ORCHESTRATION

AND ANALYSIS

VIRTUAL
DESIGN DEVELOP INTEGRATE VALIDATE

15
© by FEV.io – all rights reserved. Confidential – no passing on to third parties
DIGITAL LOOP
ACCELERATES
HOMOLOGATION

SY T
ST E N
M I NT EME
EGRATION & DATA MANAG
PR CLOU N
OJ D INFRASTRUCTURE TIO
E CT Z A
MANA NI
GEMENT OFFICE & ORGA
EXT E
ERNA
L AUDIT & COMPLIANC
FEV follows a joint technology initiative for core platform
development meeting the SDV requirements

PROJECT VALUE PREPOSITONS

System & component

excellence PLATFORM MULTI DOMAIN e.g
based on SOC ADAS PWT and E/E Consolidation
Infotainment

CLOUD HPC
CONNECTIVITY UPGRADE/UPDATE Function Split via
Data transfer seamless via OTA SOA
Established agile
cooperation scheme Feature Increase

REAL TIME
capabilities
SOFTWARE
SECURITY &
Proven SW Development DEFINED SAFETY
VEHICLE Compliance Automotive Fit
Framework
PLATFORM

© by FEV – all rights reserved. Confidential – no passing on to third parties