Next generation accelerated AIoT systems and applications. Pedro Trancoso. Special Session on EU Projects, co-located with Computing Frontiers 2023, Bologna, Italy, May 2023
1. K. Mika, R. Griessl, N. Kucza, F. Porrmann, M.
Kaiser, L. Tigges, J. Hagemeyer, P. Trancoso,
M.Waqar, F. Qararyah, S. Zouzoula, J. Ménétrey,
M. Pasin, V. Schiavoni, P. Felber, C. Marcus, O.
Brunnegard, O. Eriksson, H. Salomonsson, D.
Ödman, A. Ask, A. Casimiro, A. Bessani, T.
Carvalho, K. Gugala, P. Zierhoffer, G. Latosinski,
M. Tassemeier, M. Porrmann, H.-M. Heyn, E.
Knauss, Y. Mao, F. Meierhöfer
Next generation accelerated AIoT
systems and applications
10. May 2023
13. 13
Optimizing DL models
Harware-aware optimizations
Model compression without loss of accuracy
Optimizing Toolchain for Heterogenenous
Hardware
14. 14
Optimizing DL models
Harware-aware optimizations
Model compression without loss of accuracy
Hardware software co-design
Reconfigurable (FPGA) accelerators
Template-based description
Heterogenenous engines
Optimizing Toolchain for Heterogenenous
Hardware
Co-design
15. 15
Optimizing DL models
Harware-aware optimizations
Model compression without loss of accuracy
Hardware software co-design
Reconfigurable (FPGA) accelerators
Template-based description
Heterogenenous engines
Model verification
Load, optimize, deploy, and evaluate deep
neural networks on target hardware in a
traceable and reproducible manner.
Optimizing Toolchain for Heterogenenous
Hardware
Co-design
16. 16
Optimizing DL models
Harware-aware optimizations
Model compression without loss of accuracy
Hardware software co-design
Reconfigurable (FPGA) accelerators
Template-based description
Heterogenenous engines
Model verification
Load, optimize, deploy, and evaluate deep
neural networks on target hardware in a
traceable and reproducible manner
Simulation
Simulation framework for implementation,
testing, and debugging
Optimizing Toolchain for Heterogenenous
Hardware
Co-design
17. 17
Requirements concepts for AIoT
Conceptual model of the compositional architecture
Framework
Abstraction levels: Knowledge and Analytical, Conceptual,
Design, and Runtime level
Clusters of concerns for quality aspects of an AI system in the IoT
are Safety, Security, Privacy and Ethical Aspects
Safety aspects
Requirements Engineering facilitates constructive design of
safety critical systems
Trusted execution for AIoT
Trusted Execution Environments (TEEs): Intel SGX & ARM
TrustZone
Trusted runtimes: TWINE for SGX and WATZ for TrustZone
Inferring trust in AIoT
Remote Attestation (RA) integrated into trusted
WebAssembly runtimes
SIRE: a Byzantine fault-tolerant infrastructure supporting
remote attestation
Safety, Security and Requirements for Distributed AIoT
Systems
Group of
concerns
Cluster of
concern
Architecture
view
Architecture
viewpoint
Business Goal
or Use Case
Stakeholder
Relation /
Correspon-
dence
Level of
Abstraction
System-
of-interest
1..*
1..*
1..*
1..*
has
1..*
exists on
describes /
specifies
determines
addresses
governs
1..*
18. 18
Automotive
Pedestrian Automatic Emergency Breaking (P-AEB)
The main challenge is the limited processing power available
in the vehicle, the communication resource limits and the
possible processing power available in the edge (base
station)
Distributing the DL model over multiple processing nodes
VEDLIoT Applications
19. 19
Industrial IoT
Predictive maintenance: Motor Condition Classification
Anomaly detection: Series Arc Fault Detection (AFD) in low-voltage direct current (LVDC) systems
Challenges: limited data of fault condition under real scenarios due to finance and safety considerations,
and the lack of framework for software development with DL.
VEDLIoT Applications
20. 20
Smart home
Smart mirror - intuitive user interface, display of
personalized information and the status of the
smart home
Focus on data protection through local
processing
Tasks:
user identification by facial recognition and
tracking using depth imaging cameras
hand gestures recognized to control the mirror
voice assistant - natural language processing
(NLP) - to control key features through voice
commands
Currently deployed on one NVIDIA Orin AGX,
combined with an M.2-based Hailo-8 DL
accelerator for gesture detection
VEDLIoT Applications
RGB-D Camara
Background Removal
RetinaFace
for Face Detection
Gesture Detection
with YoloV7-tiny
Object Detection
with YoloV7
Siamese Network for
Identification
Speech to Text
with coqui
Intent Extraction
for Skill System
ROS2 Bridge
Webstream
Person Recognition Decision Maker
Hailo8
ROS2 Nodes
Text to Speech
with coqui
Micophone Input
Speaker
Trusted
Execution
Environment
MagicMirror²
Nvidia AGX Orin
21. 21
Integration of Deep Learning into IoT devices with restricted computing capabilities and
minimal power consumption requirements - energy-efficient computing
AIoT hardware platform with tailored hardware components and accelerators: from
embedded systems to edge computing and cloud platforms
Efficient middleware simplifies neural network programming, testing, and deployment to
this diverse hardware ecosystem
Innovative approaches for requirements engineering, combined with safety and security
principles
Concepts validated through use cases in vital industry sectors: automotive, automation,
and smart home
Summary