Upload
others
View
8
Download
0
Embed Size (px)
Citation preview
Smart Automotive Domain Gateways
Khaldoun Albarazi
Market Development Engineer
Automotive Product Division
STMicroelectronics
Agenda
• Automotive Industry Transformation
• Technology Drivers
• Architecture Evolution
• Advanced Processing Demand
2
Agenda
• Automotive Industry Transformation
• Technology Drivers
• Architecture Evolution
• Advanced Processing Demand
3
Automotive Industry TransformationFrom Simple Mechanics to Complex Electronics
4
Computational
Power
Electronics era
Mechanical era
Software in-car networking era
End-to-End User Experience
Environmentally Friendly
Cloud / Data Business Model
Car Engine
Gasoline / Diesel ECUs
Data Processing
Hybrid Engine
Connectivity
& Full Electrification
Poor Drivability
High maintenance costs
Electronic Engine Control
Manual Gearbox
Automatic Gearbox
Driver Assisted
Smooth Electric Engine
No gearbox / No thermal engine
Maintenance free
Mainly Analog &
Power ICs1st Simple MCU’s 16-bit MCU’s 32-bit MCU’s
Multicore, Virtual Cores, High processing,
Security & Fast connectivity
1900 1970 1990 2010 Today
• 70-90 ECUs on average
• 8-12 communication buses
• Various communication protocols
• CAN
• LIN
• Ethernet
• Flexray ..
Today’s Automotive System 5
Application Segmentation
1% 3% 4%10%
15%20%
30%35%
50%
0%
10%
20%
30%
40%
50%
60%
1950 1960 1970 1980 1990 2000 2010 2020 2030
Sha
re o
f Tota
l C
ost
Growth Automotive ECU Cost
Source: Statista
Previous & Current Architectures 6
ECU
ECU
ECU
ECU
ECU
CAN MOST
FL
EX
RA
Y
Gateway
ECU
ECU
ECU
ECU
ECU
ECU
ECU
ECU
ECU
ECU
LIN
ECU
ECU
ECU
ECU
ECU
ECU
ECU
ECU ECU
ECU ECU
Switch
ETHERNET
FLEXRAY
ECU
ECU
ECU
ECU
ECU
ECU
ECU
ECU
ECU
ECU
ECU
CAN
CA
N F
D
OB
D
Central Gateway
Legacy Network
Distributed / modular flat
architecture
Dominated by classic CAN
Low protection level
Gateway integrated into
BCM
Each ECU serves specific
function
Application-specific buses
and many 1-to-1 direct
communication links
CAN, LIN, FlexRay, MOST
Topology and Key Properties
Basic functional integration
Multiple network interfaces
More function-specific
ECUs
Security becomes crutial
Current Network
Introduction of OTA
Application-specific buses
and multiple 1-to-1 direct
communication links
CAN FD, LIN, FlexRay,
Ethernet 100Mb/s
Agenda
• Automotive Industry Transformation
• Technology Drivers
• Architecture Evolution
• Advanced Processing Demand
7
Technology Drivers 8
Major Trends
ADAS & AD
• Sensor fusion
• Safety & security
• Artificial intelligence
ELECTRIFICATION
• System integration
• Functional safety
• Cost optimization
CONNECTIVITY
• V2X
• Cloud computing
• Media streaming
• Over-The-Air
ADAS / AD 9
System, Trends, and Challenges
$330
$810
$1,243
Average Vision Radar Fusion Other ADASLevel 2-3
ADASLevel 4-5
Silicon Content ($)
Source: Strategy Analytics
System
• ADAS: Camera, Radar, and LiDAR coupled with V2X, Telematics, and GNSS
• AD: Higher level of autonomy and controllability
Trends
• ADAS value per vehicle will grow from $160 to $512
• Silicon content will grow with automation level
Challenges
• Higher level of integration
• Centralized decision making
• Safety pushed to the next level “Fail-Operational”
• Improved reaction time
• Security becomes a top safety priority
Electrification 10
System, Trends, and Challenges
EV Project Sales
Source: Bloomberg New Energy Finance
System• Hybrid Electric Vehicle (HEV):
engine plus motor (small motor/battery)
Plug-in Hybrid Electric Vehicle (PHEV):
HEV plus charger (medium motor/battery)
Battery Electric Vehicle (BEV):
PHEV minus engine (large motor/battery)
Trends• EVs will cost same as non-EVs by 2025
EVs sales will surpass non-EVs by 2038
Challenges• Functional integration of multiple
applications
• Higher safety
• Standardization
• Cost improvement
• Modular power architecture
Connectivity 11
Trends and Challenges
Source: Dr. Kirsten Matheus, BMW AG, Automotive Ethernet Congress 2019
Challenges• Higher bandwidth
• Cybersecurity
• SW and HW Standardization
• Data
• Virtualization
• Low latency
• OTA
• Adaptive Autosar / service oriented
Agenda
• Automotive Industry Transformation
• Technology Drivers
• Architecture Evolution
• Advanced Processing Demand
12
Architecture Evolotion 13
New Car Digital Architecture
Automated
Cars
Digitalization
Electrified
Cars
Connected
Cars
• Introduction of Gateways between traditional
ECUs of associated domain and backbone
• Protocol translation between signal-based and
service oriented domains with inner and cross
domain communications
• Centralization and integration of related SW
functionalities
Functional Consolidation
Domain Controller Architecture 14
ECU ECU
ECU ECU
SwitchECU
ECU
ECU
ECU
ECU
ECU
ECU
ETHERNET (SOME /IP)
EC
U
Sw
i tc
h
Domain
Chassis
Domain
Infotainment
Domain
Powertrain
ECU
ECU
ECU
ECU
ECU
ECU
ECU
ECU
ECU
ECU
ECU
ECU
Domain
Body
CA
NF
D
CA
NF
D
CA
NF
D
ET
H
OB
D
LTE
ECU
ECU
ECU
ECU
ECU
CAN MOST
FL
EX
RA
Y
Gateway
ECU
ECU
ECU
ECU
ECU
ECU
ECU
ECU
ECU
ECU
LIN
ECU
ECU
ECU
ECU
ECU
ECU
ECU
ECU ECU
ECU ECU
Switch
ETHERNET
FLEXRAY
ECU
ECU
ECU
ECU
ECU
ECU
ECU
ECU
ECU
ECU
ECU
CAN
CA
N F
D
OB
D
Central Gateway
Topology and Key Properties
Legacy
Current
Upcoming Network
• Centralization and domain fusion
with functional consolidation
• Hierarchical architecture
• High computing platforms with
smart actuators/sensors
• Many-to-1 communication with
Ethernet backbone
• CAN FD, Ethernet 1-10Gb/s
• Increased security and Intrusion
Detection Systems
• OTA everywhere
• Service oriented approach
Future Architectures 15ECU
ECU
ECU
ECU
ECU
CAN MOST
FL
EX
RA
Y
Gateway
ECU
ECU
ECU
ECU
ECU
ECU
ECU
ECU
ECU
ECU
LIN
ECU
ECU
ECU
ECU
ECU
ECU
ECU
ECU ECU
ECU ECU
Switch
ETHERNET
FLEXRAY
ECU
ECU
ECU
ECU
ECU
ECU
ECU
ECU
ECU
ECU
ECU
CAN
CA
N F
D
OB
D
Central Gateway
Topology and Key Properties
LegacyCurrent
Upcoming
Future
Future+Zone B
Central
Compute
Platform
Zone A
S /
A
S /
A
S /
A
S /
A
CAN FD
Zone CZone D
S /
A
S /
A
S /
A
S /
A
CAN FD
S /
A
S /
A
S /
A
S /
A
CAN FD
S /
A
S /
A
S /
A
S /
A
CAN FD
Central
Compute
Platform
S /
A
S /
A
S /
A
S /
A
S /
A
S /
A
S /
A
S /
A
S /
A
S /
A
S /
A
S /
A
S /
A
S /
A
S /
A
S /
A
S /
A
Future Networks
• High-level of centralization
• Domain independent with central
super computing platform
• Physical distribution and
software-driven
• All-to-some communication
• Ethernet up to 50Gb/s
Agenda
• Automotive Industry Transformation
• Technology Drivers
• Architecture Evolution
• Advanced Processing Demand
16
Advanced Processing DemandComputing and Safety Requirements 17
Safe Computing MCUSuper Computing MPU
Safety Critical System
(AD and ADAS)
Number Cruncher
Mission Planning
Behaivior Generation
Motion Planning
Probabilistic Algorithms
(Anaylsis Scenario, Make Contxtual
Decision)
Safety Processor
Sensor Processing
Dynamic Actuation
Real Time Date Managment
Deterministic Algorithms
(Initiate safe measure, FailSafe
and Fail-Operational)
Impact on MCU Design 18
Key Influncers
Micro Design &
Architetcure
CAN / ETH Routing & Inspection
Real-Time Performance
Over-The-Air
Virtualization
Safety & Security
Low Power Modes
CAN / ETH Routing &
Inspection Routing performance
Packet inspection
Software overload
New protocols
Real Time Behavior Start-up / power-down
Reaction time
Accelerators
Over The-Air Hardware support
Rollback functionality
External Memory interfaces
Secure boot
Virtualization HW hypervisor
Real-time, determistic
behvior
Safety & Security ASIL-D support
HW self-test
HSM subsytsem
Memory protection
Communication secuirty
Low Power Modes Standby/Stop operations
Smart wakeup unit
1st Automotive MCU
w. real-time Virtualization
ie. Multi Update-able
Applications Platform
(turn-key for DCUs)
Unprecedented
Safe & Secure
Real-time
Performance
STELLAR MCU Family 19
Unique Solution in 28nm eNVM Technology Paradigm
CMOS
28nm FD-SOI
PCM
Embedded NVM
Phase Change Memory
ARM real-time CPU
STELLAR Real-Time
HW Virtualization Architecture built on
only real-time CPU w.
HW virtualization
most powerful
real-time CPU +
STELLAR MCU Family 20
Processing Architecture - Heterogeneous Multiprocessor
• Safe Data Router
• Safe / QM Input/Ouput
Data Processing &
Move for Higher Level
Proc. or Output
• Security Subsystem
1 Gbps Ethernet w. TSN
SENT, PSI5/PSI5-S,
MSC+
Hyperbus / Octal SPI
eMMC
Zipwire
PCIe 3.0In
terf
ace
s
M4 Subsystems Use Cases
CAN-FD, LIN-UART,
FlexRay, SPI, I2C
Lock-Step
Cortex-R52
400MHz
I-/D-Cache
TCM
NEON
Lock-Step
Cortex-R52
400MHz
TCM
I-/D-Cache
RAMPCM
Lock-Step
Cortex-R52
400MHz
I-/D-Cache
TCM
NEON
Lock-Step
Cortex-R52
400MHz
TCM
I-/D-Cache
Split-Lock
Cortex-R52
400MHz
TCM
Cortex-R52
400MHz
TCM
I-/D-Cache
I-/D-Cache
Lock-Step
Cortex-M4
200MHz
RAM
Cortex-M4
200MHz
RAMPCM
Crypto CoreAES/PKA/SHA/TRNG
Lock-Step
AES Crypto
Cortex-M4
200MHz
RAM
Data Move / Processing
data PCM
SAFE NETWORK ON CHIP
DM
As
RAM
Memory
• Up to 20M PCM
• Up to 8.5M RAM
RAMPCM
RAMPCM
...
Core
RAM
...
Timer Modules
GTM4
Ultra Real-time Input / Output (Pre)-Processing Real-time Ultra Performance
...
...
...
12-b
itS
AR
AD
C
16-b
it
SD
AD
C
ADC
DSP
25
0n
s A
na
log
Com
p
SIM
D A
cc.
Security Island
Power
Management
Thank You!
21
Contact
Khaldoun AlbaraziMarket Development Engineer
Automotive Product Division
STMicroelectronics