SlideShare a Scribd company logo
Evolution and Trends in Edge AI Systems and
Architectures for the Internet of Things Era
Prof. David Atienza Alonso
Embedded Systems Laboratory (ESL), EPFL
david.atienza@epfl.ch
FDI-Conferencia de Posgrado, UCM, 17 de noviembre, 2021
https://www.epfl.ch/labs/esl/
© EPFL 2021
 Team covers IoT and edge AI (both hardware and software co-design)
 Headed by Prof. David Atienza
 All started in 2008, 45 members today:
 8 post-docs
 22 PhD students (18 grad)
 11 M.Sc. Theses (26 grad)
 2 senior engineers
 1 administrative assistant, 1 system admin.
 Support from academia and industry for R&D projects
 31 companies/R&D centers provide equipment, grants and donations
David Atienza
2
ESL Team and Collaborations
© EPFL 2021
Origin of IoT: Wireless Sensor Networks
 Late 80s: Wireless Sensor Networks (WSNs) are data acquisition and distribution
networks consisting of a set of (large) sensor nodes
 Simple architectures (“get” data)
 Cooperative applications and transmission
David Atienza
1
2
Sink
3
4
T = 26.85 ºC
T = 31.85 ºC
T = 24.85 ºC
T = 24.85 ºC is OK
© EPFL 2021
In 2012 - The Internet of Things Concept :
“Nano Sensors with Good Connectivity”
David Atienza
 Few more years beyond 1999, but thanks to Moore’s Law, after 45 years…
Complexity increase
Intel 4004 [1971]
(2250 p-MOS trans.,108 KHz)
Busicon Calculator 141-PF
Intel 5th Gen Core i7 vPro [2012]
(1.3B trans., 3.3GHz)
2012
1971
© EPFL 2021
Wearables
Connected
Cars
Connected City
Industry 4.0
Transportation
Healthcare
Oil & Gas
[Source: DARPA and EC]
Continuous system
monitoring
(IoT sensors)
David Atienza
IoT System-Level Architecture
Big-Data Cloud
computing
Connected
Houses
© EPFL 2021 David Atienza
IoT Vision: Guardian Angels
See more details at: https://www.ga-project.eu/
© EPFL 2021
IoT Enabling Digital Era: Technology Convergence
 IoT concept is the convergence of three technologies within “Digital Era”
+ Digitized Enterprise data
1
IoT sensors
(Internet of Everything)
Analytics and
Deep Learning
3
Big data storage and
Computing IoT platforms
2
Able to collect
unprecedented
amounts of data
Capacity to store
and handle large
amounts of data
Capacity to make
sense of this data
in a confident way
David Atienza
© EPFL 2021
IoT (Potential) Benefits
David Atienza
 IoT long-term economic benefits [McKinsey]
 Remote healthcare: $11.1Trill/year saved (1B people)
 Efficient energy: 45TWh/year saved in EU (4M homes)
 Automotive industry supply chain: 50% costs reduction
 …
Business-to-business services: 70% added value! But…
How can IoT be designed?
© EPFL 2021
 Two-fold paradigm shift in health delivery
 Cardiovascular monitoring is key today…
Environment
Genetics
Access to care
Health
behaviors,
personal
lifestyle
Determinants of health issues
(source: Institute for the future, Center for
disease control and prevention, 2006)
Symptom-based  Preventive healthcare
Hospital-centered  Person-centered
Trainer/coach
Home record
ECG Holter data logger
(clinical practice)
Resting Electrocardiogram
(ECG)
IoT to the Rescue of Our Healthcare System
David Atienza
© EPFL 2021
Apple Watch System: Example of (Expensive) IoT Sensor
 April 2015: Combination of Watch + iPhone
 IoT system design
 Watch: “Raw” bio-signals acquisition
 iPhone: User interface and long-range transmission
 Cloud: Bio-signal (post-)processing
David Atienza
iPhone 6: A8 MPCore
(Dual-core CPU, 4-core GPU)
iPhone was less powerful!
(ARM 1176JZF-S)
Apple Watch: APL 0778 processor
(with 4GB, 5 sensors, loudspeaker, …)
Limited lifetime (< 5 hours for bio-signal analysis) and no real-time…
COMMUNICATION OVERHEAD!
© EPFL 2021
ECG
Sensing and
sampling
Processing
Communication
Energy
consumption
breakdown
 This ECG streaming monitor lasts 5h, why?
 Send 16 bits: more energy than doing 64’000 operations!
Long-Lived IoT Designs Feasible Today?
David Atienza
IoT age requires complete system revolution!
How to reduce IoT data sensed and streamed?
Only get and keep “useful” data from IoT sensors!
© EPFL 2021
So, Is IoT Still on “Top of the Wave”?
 Very high expectations since 2014…
[Source: Gartner]
David Atienza
Did the “IoT dream” disappear?!
© EPFL 2021
IoT just got “reborn”…
[Source: Gartner]
New IoT revolution, 3 technologies:
(1) Edge AI
(2) IoT platform
(3) Deep learning (learn from Big Data!)
David Atienza
© EPFL 2021
New IoT Platform (“Smart Cloud”): 3 Delivery Modes
 Infrastructure as a Service (IaaS): IT equipment
 Platform as a Service (PaaS): Big Data access and DL libraries
 Software as a Service (SaaS): Full AI application services
Platform as a
Service (PaaS)
SaaS
PaaS
IaaS
$
$$$$$
David Atienza
© EPFL 2021
What is “Edge AI / Computing”?
David Atienza 15
 Intelligence is moving towards edge devices
 Real-time pre-filtering and data processing
 IoT devices send only a portion of the data to the
central data repository, or corporate datacenter
 Reliable results of Big Data AI on IoT Cloud platforms
© EPFL 2021 David Atienza
Does Data Size Matters for a Global IoT Business Perspective?
3x cost reduction if data
volume is reduced by 95%
Yes, only send (and keep) useful data: powerful IoT edge AI platforms
© EPFL 2021
IoT Edge AI Concept: Complex Hardware Template
David Atienza
RF + COMPUTING
© EPFL 2021
New Edge AI + Patches Can Understand Human Movements
 New collaborative work with multiple patches and one IoT edge AI node
 Multiple patches to track in real-time: but AI calibration needed
David Atienza
© EPFL 2021
Can you Use the edge AI Concept for new Businesses: Can you be a Ski Coach?
Just Add An App and Earbuds!
 Personalized coach with
IoT sensors for ski
 Personalized App
 Create metrics based
upon 4 fundamental skiing
skills:
 Edging,
 Balance,
 Rotation,
 Pressure.
David Atienza
Based on few types of sensors and calibrated AI on the
App, do you get something you can sell?
© EPFL 2021 David Atienza
See more details at: https://getcarv.com/
Can you Use the edge AI Concept for new Businesses: Can you be a Ski Coach?
Just An App and Earbuds!
© EPFL 2021
IoT Edge AI Sensing: Few Sensors for Many Application Domains
 Sensors specific per application: but large convergence in similar products
David Atienza
© EPFL 2021
 Highly-integrated Apple S6 SiP Chip
 ARM DualCore SC300 @ 780 MHz
 Integrated graphics and PowerVR (ML accel.)
 32 GB of memory
 Multiple sensors
 GPS + GLONASS geolocat.
 Heart Monitor sensor: PPG-based HR
 Electrical heart sensor
 3-axis accelerometer + gyroscope
 Ambient light sensor, Built-in compass
 Multiple radios included (+ eSIM)
 Gold antenna embedded in casing
 LTE/UMTS, NFC, Wi-Fi 802.11b/g/n, BT 5.0
 Price: $399
Example (new) IoT Edge AI for Healthcare: Apple Watch Series 6
David Atienza
New IoT Edge AI wearables (with limited set of sensors)
provide minimal latency and more lifetime
© EPFL 2021
What can new Edge AI Sensors for Healthcare do?
 Same sensors, but algorithms and performance
improved with edge AI accelerator chips
David Atienza
Accurate detection of cardiovascular pathologies
without hospitals and specialized personnel!
© EPFL 2021
Example IoT Edge AI for Smart Home Devices: “Personalized” Coffee
 New IoT products already on the market… Control Edge AI, Big Data and DL
IoT Edge AI saves energy and gives personalization:
Is then connectivity still relevant?
David Atienza
Custom IoT Edge
AI design: $3
© EPFL 2021
New Open-Source Custom Hardware of ULP SoCs for IoT Edge AI Devices
 Case Study: GAP8 Multi-Core SoC Edge Platform
 RISC-V architecture (Based on ETHZ PULP)
Source: GAP8 SDK internal documentation
(https://github.com/GreenWaves-Technologies/gap_sdk)
1 general
purpose
core
8-core cluster for complex
signal processing
A lot of
memory
28nm FD-SOI
New Deep
Learning
accelerators
Custom IoT Edge
AI design: $3
© EPFL 2021
Vertical Companies Created to Control New IoT Space
 IoT business Landscape: Data-related services more profitable
David Atienza
Source: Yole Dev. ’17 (Upd.)
© EPFL 2021
 IoT business Landscape: Data-related services more profitable
 But largest companies trying to avoid risks: full chain control
David Atienza
Vertical Companies Created to Control New IoT Space
Amazon Echo
Amazon Zero Touch
© EPFL 2021
New Extended Support in Cloud for Secure IoT Devices:
Example of Amazon Web Services (AWS IoT) and Google Teachable Machines
 Open-source IoT communication standards supported: Message Queuing Telemetry Transport (MQTT)
 Enable new IoT persistence concept: Device Shadow concept (Virtual Twin) when disconnected
(Official)
Secured
devices
David Atienza
© EPFL 2021
New IoT Revolution… Enabling Industry 4.0!
… Collect data from
everywhere
… Exploit all sources
to make decisions
… New ways to interact
with machines
… New robotics help
deliver efficiency
Monitoring (all) and
maintenance
Virtual twins and
model predictions
Human-Machine
interfaces
Digital autonomy in
physical world
Edge IoT
Big Data analytics
Wearables
Augmented reality
Advanced robotics
Deep learning
Deep learning
Edge IoT
Edge IoT
Deep learning
Edge IoT
David Atienza
© EPFL 2021
New Robotics Systems for IoT: Edge AI + Sensing Evolution for Real-Time Reaction
 From frame- to event-based Dynamic Vision System (DVS): analog frontend off most of the time
Frame-
based
{x
0
, y
0
}
{x
1
, y
1
}
{x
2
, y
2
}
{x
3
, y
3
}
{x
N-1
, y
N-1
}
Event-
based
David Atienza
PULPv3
GrainCam
Mixed-Signal
Event-based
Imager
Digital
Parallel
Processor
[Courtesy: L. Benini]
© EPFL 2021
 Insightness: Obstacle avoidance with 7-layer Deep neural network
 Learn with public traffic databases
Delivery with Drones: React in Real-Time with IoT Edge + Deep Learning
[Courtesy: Davide Scaramuzza, RPL-Univ. Zurich]
David Atienza
© EPFL 2021
IoT just got “reborn”… But keeps evolving in 2021!
David Atienza
[Source: Gartner]
But 3 techs. of IoT revolution remain:
(1) Edge AI: Embedded AI, carbon-based transistors,
biodegradable sensors, ...
(2) IoT platforms: SASE, data fabric, digital twin of the
person, citizen twin, composable infrastructure or
enterprise, ...
(3) Deep learning/AI: Explainable AI, Responsible AI,
Generative AI, Composite AI, Adaptive ML, GANs,..
(4) Communication key again: 5G?
© EPFL 2021
New 5G Connectivity for IoT: Between Local vs Global Networks
Today: Local Networks (Legacy)
 Simple local networks, short range links
(scalable so far, but with latency)
 Requires local infrastructure devices,
setup, and maintenance
 Transceiver simple/cheap
 No extra service subscription (low rates)
 Reuse of existing comm. resources
Global Connectivity (5G)
 Complex (cellular) network with long range
links
 No local infrastructure with some QoS by
service provider
 Complex/costly transceiver
 Requires service subscription
 Limited reuse, no shared comm. resources
David Atienza
Evolution promised/envisioned by 5G
[Courtesy: Andreas Burg (TCL@EPFL)]
© EPFL 2021
Conclusion
 New IoT concept possible due to techs. convergence: real-time response, safer and better AI
 Upcoming technology trends:
 Even more distributed/hierarchical intelligence in future IoT Era
 “Copy” more from biology (More specialization on processors = energy efficiency, etc.),
 Finding schemes to secure and activate selectively IoT devices (Self-awareness)
1 Edge AI sensors
(and actuators)
Analytics and Deep
Learning
3
Big data storage and
Computing IoT platforms
2
Unprecedented amounts of
(useful) data at the edge/fog
Capacity to store and
handle large data volumes
Capacity to make sense
of it (automatically)
David Atienza
© EPFL 2021
QUESTIONS?
david.atienza@epfl.ch
Thank You!
David Atienza

More Related Content

What's hot

Top 5 Deep Learning and AI Stories - November 3, 2017
Top 5 Deep Learning and AI Stories - November 3, 2017Top 5 Deep Learning and AI Stories - November 3, 2017
Top 5 Deep Learning and AI Stories - November 3, 2017
NVIDIA
 
GTC China 2017 Highlights
GTC China 2017 HighlightsGTC China 2017 Highlights
GTC China 2017 Highlights
NVIDIA
 
Cloud Computing, SDN, Big Data and Internet of Everything - Lew Tucker
Cloud Computing, SDN, Big Data and Internet of Everything - Lew TuckerCloud Computing, SDN, Big Data and Internet of Everything - Lew Tucker
Cloud Computing, SDN, Big Data and Internet of Everything - Lew Tucker
Lew Tucker
 
Machine learning and ai in a brave new cloud world
Machine learning and ai in a brave new cloud worldMachine learning and ai in a brave new cloud world
Machine learning and ai in a brave new cloud world
Ulf Mattsson
 
Wiki stage 20151128 - v001
Wiki stage   20151128 - v001Wiki stage   20151128 - v001
Wiki stage 20151128 - v001
Enrico Busto
 
Machine Learning Applications to IoT
Machine Learning Applications to IoTMachine Learning Applications to IoT
Machine Learning Applications to IoT
Förderverein Technische Fakultät
 
Internet of Things (ruSMART 2013, St. Petersburg)
Internet of Things (ruSMART 2013, St. Petersburg)Internet of Things (ruSMART 2013, St. Petersburg)
Internet of Things (ruSMART 2013, St. Petersburg)
Rob Van Den Dam
 
Ai open powermeetupmarch25th
Ai open powermeetupmarch25thAi open powermeetupmarch25th
Ai open powermeetupmarch25th
IBM
 
Kush stats alpha
Kush stats alpha Kush stats alpha
Kush stats alpha
Ajay Ohri
 
Conversational Architecture, CAVE Language, Data Stewardship
Conversational Architecture, CAVE Language, Data StewardshipConversational Architecture, CAVE Language, Data Stewardship
Conversational Architecture, CAVE Language, Data Stewardship
Loren Davie
 
The Next Generation of AI and Deep Learning - GTC17
The Next Generation of AI and Deep Learning - GTC17The Next Generation of AI and Deep Learning - GTC17
The Next Generation of AI and Deep Learning - GTC17
NVIDIA
 
Leveraging open source for big data stack
Leveraging open source for big data stackLeveraging open source for big data stack
Leveraging open source for big data stack
Flytxt
 
Architectural cncepts: Chip Multithreaded Era
Architectural cncepts: Chip Multithreaded EraArchitectural cncepts: Chip Multithreaded Era
Architectural cncepts: Chip Multithreaded Era
Marcelo Arbore
 
Ibm big data-platform
Ibm big data-platformIbm big data-platform
Ibm big data-platform
IBM Sverige
 
NextGen Infrastructure for Big Data
NextGen Infrastructure for Big DataNextGen Infrastructure for Big Data
NextGen Infrastructure for Big Data
Ed Dodds
 
EDW 2015 cognitive computing panel session
EDW 2015 cognitive computing panel session EDW 2015 cognitive computing panel session
EDW 2015 cognitive computing panel session
Steve Ardire
 
A Methodology for Building the Internet of Things
A Methodology for Building the Internet of ThingsA Methodology for Building the Internet of Things
A Methodology for Building the Internet of Things
The Internet of Things Methodology
 
Artificial Intelligence and Cognitive Computing
Artificial Intelligence and Cognitive ComputingArtificial Intelligence and Cognitive Computing
Artificial Intelligence and Cognitive Computing
Florian Georg
 
re:Invent re:Cap - Big Data & IoT at Any Scale
re:Invent re:Cap - Big Data & IoT at Any Scalere:Invent re:Cap - Big Data & IoT at Any Scale
re:Invent re:Cap - Big Data & IoT at Any Scale
Adrian Hornsby
 
14 Startups Leading the Artificial Intelligence (AI) Revolution
14 Startups Leading the Artificial Intelligence (AI) Revolution14 Startups Leading the Artificial Intelligence (AI) Revolution
14 Startups Leading the Artificial Intelligence (AI) Revolution
NVIDIA
 

What's hot (20)

Top 5 Deep Learning and AI Stories - November 3, 2017
Top 5 Deep Learning and AI Stories - November 3, 2017Top 5 Deep Learning and AI Stories - November 3, 2017
Top 5 Deep Learning and AI Stories - November 3, 2017
 
GTC China 2017 Highlights
GTC China 2017 HighlightsGTC China 2017 Highlights
GTC China 2017 Highlights
 
Cloud Computing, SDN, Big Data and Internet of Everything - Lew Tucker
Cloud Computing, SDN, Big Data and Internet of Everything - Lew TuckerCloud Computing, SDN, Big Data and Internet of Everything - Lew Tucker
Cloud Computing, SDN, Big Data and Internet of Everything - Lew Tucker
 
Machine learning and ai in a brave new cloud world
Machine learning and ai in a brave new cloud worldMachine learning and ai in a brave new cloud world
Machine learning and ai in a brave new cloud world
 
Wiki stage 20151128 - v001
Wiki stage   20151128 - v001Wiki stage   20151128 - v001
Wiki stage 20151128 - v001
 
Machine Learning Applications to IoT
Machine Learning Applications to IoTMachine Learning Applications to IoT
Machine Learning Applications to IoT
 
Internet of Things (ruSMART 2013, St. Petersburg)
Internet of Things (ruSMART 2013, St. Petersburg)Internet of Things (ruSMART 2013, St. Petersburg)
Internet of Things (ruSMART 2013, St. Petersburg)
 
Ai open powermeetupmarch25th
Ai open powermeetupmarch25thAi open powermeetupmarch25th
Ai open powermeetupmarch25th
 
Kush stats alpha
Kush stats alpha Kush stats alpha
Kush stats alpha
 
Conversational Architecture, CAVE Language, Data Stewardship
Conversational Architecture, CAVE Language, Data StewardshipConversational Architecture, CAVE Language, Data Stewardship
Conversational Architecture, CAVE Language, Data Stewardship
 
The Next Generation of AI and Deep Learning - GTC17
The Next Generation of AI and Deep Learning - GTC17The Next Generation of AI and Deep Learning - GTC17
The Next Generation of AI and Deep Learning - GTC17
 
Leveraging open source for big data stack
Leveraging open source for big data stackLeveraging open source for big data stack
Leveraging open source for big data stack
 
Architectural cncepts: Chip Multithreaded Era
Architectural cncepts: Chip Multithreaded EraArchitectural cncepts: Chip Multithreaded Era
Architectural cncepts: Chip Multithreaded Era
 
Ibm big data-platform
Ibm big data-platformIbm big data-platform
Ibm big data-platform
 
NextGen Infrastructure for Big Data
NextGen Infrastructure for Big DataNextGen Infrastructure for Big Data
NextGen Infrastructure for Big Data
 
EDW 2015 cognitive computing panel session
EDW 2015 cognitive computing panel session EDW 2015 cognitive computing panel session
EDW 2015 cognitive computing panel session
 
A Methodology for Building the Internet of Things
A Methodology for Building the Internet of ThingsA Methodology for Building the Internet of Things
A Methodology for Building the Internet of Things
 
Artificial Intelligence and Cognitive Computing
Artificial Intelligence and Cognitive ComputingArtificial Intelligence and Cognitive Computing
Artificial Intelligence and Cognitive Computing
 
re:Invent re:Cap - Big Data & IoT at Any Scale
re:Invent re:Cap - Big Data & IoT at Any Scalere:Invent re:Cap - Big Data & IoT at Any Scale
re:Invent re:Cap - Big Data & IoT at Any Scale
 
14 Startups Leading the Artificial Intelligence (AI) Revolution
14 Startups Leading the Artificial Intelligence (AI) Revolution14 Startups Leading the Artificial Intelligence (AI) Revolution
14 Startups Leading the Artificial Intelligence (AI) Revolution
 

Similar to Evolution and Trends in Edge AI Systems and Architectures for the Internet of Things Era

bhide_connected_raleigh2016 (1)
bhide_connected_raleigh2016 (1)bhide_connected_raleigh2016 (1)
bhide_connected_raleigh2016 (1)
sandhibhide
 
Developing-IoT-Solutions
Developing-IoT-SolutionsDeveloping-IoT-Solutions
Developing-IoT-Solutions
Manoj Punamia
 
iot_ppt.pptx
iot_ppt.pptxiot_ppt.pptx
iot_ppt.pptx
Eslam El Hefny
 
IOT_PPT1.pdf
IOT_PPT1.pdfIOT_PPT1.pdf
IOT_PPT1.pdf
laxmikanth45
 
2 pc enterprise summit cronin newfinal aug 18
2 pc enterprise summit cronin newfinal aug 182 pc enterprise summit cronin newfinal aug 18
2 pc enterprise summit cronin newfinal aug 18
IntelAPAC
 
What is the internet of things v3
What is the internet of things v3What is the internet of things v3
What is the internet of things v3
Incubation & Industry
 
Iot presentation
Iot presentationIot presentation
Iot presentation
huma742446
 
Connected Things, IoT and 5G
Connected Things, IoT and 5GConnected Things, IoT and 5G
Connected Things, IoT and 5G
Eueung Mulyana
 
IOT
IOTIOT
TheValueChain Beyond Simple 10-05-16 - Internet of Things
TheValueChain Beyond Simple 10-05-16 - Internet of ThingsTheValueChain Beyond Simple 10-05-16 - Internet of Things
TheValueChain Beyond Simple 10-05-16 - Internet of Things
TheValueChain
 
Internet of things-IoT.pptx
Internet of things-IoT.pptxInternet of things-IoT.pptx
Internet of things-IoT.pptx
Mukulislam1
 
CECA - Collaborative Approach for Decision Makers in Heterogeneous IoT Enviro...
CECA - Collaborative Approach for Decision Makers in Heterogeneous IoT Enviro...CECA - Collaborative Approach for Decision Makers in Heterogeneous IoT Enviro...
CECA - Collaborative Approach for Decision Makers in Heterogeneous IoT Enviro...
Edson Barbosa
 
Basics of Internet of Things
 Basics of Internet of Things Basics of Internet of Things
Basics of Internet of Things
Brainware University
 
10×6 Fintech: un nouveau pilier? Pascal Laffineur, Altran Benelux
10×6 Fintech: un nouveau pilier?  Pascal Laffineur, Altran Benelux10×6 Fintech: un nouveau pilier?  Pascal Laffineur, Altran Benelux
10×6 Fintech: un nouveau pilier? Pascal Laffineur, Altran Benelux
Thierry Raizer
 
Huawei empowers healthcare industry with AI technology
Huawei empowers healthcare industry with AI technologyHuawei empowers healthcare industry with AI technology
Huawei empowers healthcare industry with AI technology
Heiko Joerg Schick
 
Small Be The New Big! Let's shrink the world with enterprise IoT solutions
Small Be The New Big! Let's shrink the world with enterprise IoT solutionsSmall Be The New Big! Let's shrink the world with enterprise IoT solutions
Small Be The New Big! Let's shrink the world with enterprise IoT solutions
Tanisha Bajaj
 
The-Internet-Of-Things-4th-Industrial-Revolution.pptx
The-Internet-Of-Things-4th-Industrial-Revolution.pptxThe-Internet-Of-Things-4th-Industrial-Revolution.pptx
The-Internet-Of-Things-4th-Industrial-Revolution.pptx
HadHic
 
IEEE Computer Society Phoenix Chapter - Internet of Things Innovations & Mega...
IEEE Computer Society Phoenix Chapter - Internet of Things Innovations & Mega...IEEE Computer Society Phoenix Chapter - Internet of Things Innovations & Mega...
IEEE Computer Society Phoenix Chapter - Internet of Things Innovations & Mega...
Mark Goldstein
 
it.pptx
it.pptxit.pptx
AIoT: Intelligence on Microcontroller
AIoT: Intelligence on MicrocontrollerAIoT: Intelligence on Microcontroller
AIoT: Intelligence on Microcontroller
Andri Yadi
 

Similar to Evolution and Trends in Edge AI Systems and Architectures for the Internet of Things Era (20)

bhide_connected_raleigh2016 (1)
bhide_connected_raleigh2016 (1)bhide_connected_raleigh2016 (1)
bhide_connected_raleigh2016 (1)
 
Developing-IoT-Solutions
Developing-IoT-SolutionsDeveloping-IoT-Solutions
Developing-IoT-Solutions
 
iot_ppt.pptx
iot_ppt.pptxiot_ppt.pptx
iot_ppt.pptx
 
IOT_PPT1.pdf
IOT_PPT1.pdfIOT_PPT1.pdf
IOT_PPT1.pdf
 
2 pc enterprise summit cronin newfinal aug 18
2 pc enterprise summit cronin newfinal aug 182 pc enterprise summit cronin newfinal aug 18
2 pc enterprise summit cronin newfinal aug 18
 
What is the internet of things v3
What is the internet of things v3What is the internet of things v3
What is the internet of things v3
 
Iot presentation
Iot presentationIot presentation
Iot presentation
 
Connected Things, IoT and 5G
Connected Things, IoT and 5GConnected Things, IoT and 5G
Connected Things, IoT and 5G
 
IOT
IOTIOT
IOT
 
TheValueChain Beyond Simple 10-05-16 - Internet of Things
TheValueChain Beyond Simple 10-05-16 - Internet of ThingsTheValueChain Beyond Simple 10-05-16 - Internet of Things
TheValueChain Beyond Simple 10-05-16 - Internet of Things
 
Internet of things-IoT.pptx
Internet of things-IoT.pptxInternet of things-IoT.pptx
Internet of things-IoT.pptx
 
CECA - Collaborative Approach for Decision Makers in Heterogeneous IoT Enviro...
CECA - Collaborative Approach for Decision Makers in Heterogeneous IoT Enviro...CECA - Collaborative Approach for Decision Makers in Heterogeneous IoT Enviro...
CECA - Collaborative Approach for Decision Makers in Heterogeneous IoT Enviro...
 
Basics of Internet of Things
 Basics of Internet of Things Basics of Internet of Things
Basics of Internet of Things
 
10×6 Fintech: un nouveau pilier? Pascal Laffineur, Altran Benelux
10×6 Fintech: un nouveau pilier?  Pascal Laffineur, Altran Benelux10×6 Fintech: un nouveau pilier?  Pascal Laffineur, Altran Benelux
10×6 Fintech: un nouveau pilier? Pascal Laffineur, Altran Benelux
 
Huawei empowers healthcare industry with AI technology
Huawei empowers healthcare industry with AI technologyHuawei empowers healthcare industry with AI technology
Huawei empowers healthcare industry with AI technology
 
Small Be The New Big! Let's shrink the world with enterprise IoT solutions
Small Be The New Big! Let's shrink the world with enterprise IoT solutionsSmall Be The New Big! Let's shrink the world with enterprise IoT solutions
Small Be The New Big! Let's shrink the world with enterprise IoT solutions
 
The-Internet-Of-Things-4th-Industrial-Revolution.pptx
The-Internet-Of-Things-4th-Industrial-Revolution.pptxThe-Internet-Of-Things-4th-Industrial-Revolution.pptx
The-Internet-Of-Things-4th-Industrial-Revolution.pptx
 
IEEE Computer Society Phoenix Chapter - Internet of Things Innovations & Mega...
IEEE Computer Society Phoenix Chapter - Internet of Things Innovations & Mega...IEEE Computer Society Phoenix Chapter - Internet of Things Innovations & Mega...
IEEE Computer Society Phoenix Chapter - Internet of Things Innovations & Mega...
 
it.pptx
it.pptxit.pptx
it.pptx
 
AIoT: Intelligence on Microcontroller
AIoT: Intelligence on MicrocontrollerAIoT: Intelligence on Microcontroller
AIoT: Intelligence on Microcontroller
 

More from Facultad de Informática UCM

¿Por qué debemos seguir trabajando en álgebra lineal?
¿Por qué debemos seguir trabajando en álgebra lineal?¿Por qué debemos seguir trabajando en álgebra lineal?
¿Por qué debemos seguir trabajando en álgebra lineal?
Facultad de Informática UCM
 
TECNOPOLÍTICA Y ACTIVISMO DE DATOS: EL MAPEO COMO FORMA DE RESILIENCIA ANTE L...
TECNOPOLÍTICA Y ACTIVISMO DE DATOS: EL MAPEO COMO FORMA DE RESILIENCIA ANTE L...TECNOPOLÍTICA Y ACTIVISMO DE DATOS: EL MAPEO COMO FORMA DE RESILIENCIA ANTE L...
TECNOPOLÍTICA Y ACTIVISMO DE DATOS: EL MAPEO COMO FORMA DE RESILIENCIA ANTE L...
Facultad de Informática UCM
 
DRAC: Designing RISC-V-based Accelerators for next generation Computers
DRAC: Designing RISC-V-based Accelerators for next generation ComputersDRAC: Designing RISC-V-based Accelerators for next generation Computers
DRAC: Designing RISC-V-based Accelerators for next generation Computers
Facultad de Informática UCM
 
uElectronics ongoing activities at ESA
uElectronics ongoing activities at ESAuElectronics ongoing activities at ESA
uElectronics ongoing activities at ESA
Facultad de Informática UCM
 
Tendencias en el diseño de procesadores con arquitectura Arm
Tendencias en el diseño de procesadores con arquitectura ArmTendencias en el diseño de procesadores con arquitectura Arm
Tendencias en el diseño de procesadores con arquitectura Arm
Facultad de Informática UCM
 
Formalizing Mathematics in Lean
Formalizing Mathematics in LeanFormalizing Mathematics in Lean
Formalizing Mathematics in Lean
Facultad de Informática UCM
 
Introduction to Quantum Computing and Quantum Service Oriented Computing
Introduction to Quantum Computing and Quantum Service Oriented ComputingIntroduction to Quantum Computing and Quantum Service Oriented Computing
Introduction to Quantum Computing and Quantum Service Oriented Computing
Facultad de Informática UCM
 
Computer Design Concepts for Machine Learning
Computer Design Concepts for Machine LearningComputer Design Concepts for Machine Learning
Computer Design Concepts for Machine Learning
Facultad de Informática UCM
 
Inteligencia Artificial en la atención sanitaria del futuro
Inteligencia Artificial en la atención sanitaria del futuroInteligencia Artificial en la atención sanitaria del futuro
Inteligencia Artificial en la atención sanitaria del futuro
Facultad de Informática UCM
 
Design Automation Approaches for Real-Time Edge Computing for Science Applic...
 Design Automation Approaches for Real-Time Edge Computing for Science Applic... Design Automation Approaches for Real-Time Edge Computing for Science Applic...
Design Automation Approaches for Real-Time Edge Computing for Science Applic...
Facultad de Informática UCM
 
Estrategias de navegación para robótica móvil de campo: caso de estudio proye...
Estrategias de navegación para robótica móvil de campo: caso de estudio proye...Estrategias de navegación para robótica móvil de campo: caso de estudio proye...
Estrategias de navegación para robótica móvil de campo: caso de estudio proye...
Facultad de Informática UCM
 
Fault-tolerance Quantum computation and Quantum Error Correction
Fault-tolerance Quantum computation and Quantum Error CorrectionFault-tolerance Quantum computation and Quantum Error Correction
Fault-tolerance Quantum computation and Quantum Error Correction
Facultad de Informática UCM
 
Cómo construir un chatbot inteligente sin morir en el intento
Cómo construir un chatbot inteligente sin morir en el intentoCómo construir un chatbot inteligente sin morir en el intento
Cómo construir un chatbot inteligente sin morir en el intento
Facultad de Informática UCM
 
Automatic generation of hardware memory architectures for HPC
Automatic generation of hardware memory architectures for HPCAutomatic generation of hardware memory architectures for HPC
Automatic generation of hardware memory architectures for HPC
Facultad de Informática UCM
 
Type and proof structures for concurrency
Type and proof structures for concurrencyType and proof structures for concurrency
Type and proof structures for concurrency
Facultad de Informática UCM
 
Hardware/software security contracts: Principled foundations for building sec...
Hardware/software security contracts: Principled foundations for building sec...Hardware/software security contracts: Principled foundations for building sec...
Hardware/software security contracts: Principled foundations for building sec...
Facultad de Informática UCM
 
Jose carlossancho slidesLa seguridad en el desarrollo de software implementad...
Jose carlossancho slidesLa seguridad en el desarrollo de software implementad...Jose carlossancho slidesLa seguridad en el desarrollo de software implementad...
Jose carlossancho slidesLa seguridad en el desarrollo de software implementad...
Facultad de Informática UCM
 
Do you trust your artificial intelligence system?
Do you trust your artificial intelligence system?Do you trust your artificial intelligence system?
Do you trust your artificial intelligence system?
Facultad de Informática UCM
 
Redes neuronales y reinforcement learning. Aplicación en energía eólica.
Redes neuronales y reinforcement learning. Aplicación en energía eólica.Redes neuronales y reinforcement learning. Aplicación en energía eólica.
Redes neuronales y reinforcement learning. Aplicación en energía eólica.
Facultad de Informática UCM
 
Challenges and Opportunities for AI and Data analytics in Offshore wind
Challenges and Opportunities for AI and Data analytics in Offshore windChallenges and Opportunities for AI and Data analytics in Offshore wind
Challenges and Opportunities for AI and Data analytics in Offshore wind
Facultad de Informática UCM
 

More from Facultad de Informática UCM (20)

¿Por qué debemos seguir trabajando en álgebra lineal?
¿Por qué debemos seguir trabajando en álgebra lineal?¿Por qué debemos seguir trabajando en álgebra lineal?
¿Por qué debemos seguir trabajando en álgebra lineal?
 
TECNOPOLÍTICA Y ACTIVISMO DE DATOS: EL MAPEO COMO FORMA DE RESILIENCIA ANTE L...
TECNOPOLÍTICA Y ACTIVISMO DE DATOS: EL MAPEO COMO FORMA DE RESILIENCIA ANTE L...TECNOPOLÍTICA Y ACTIVISMO DE DATOS: EL MAPEO COMO FORMA DE RESILIENCIA ANTE L...
TECNOPOLÍTICA Y ACTIVISMO DE DATOS: EL MAPEO COMO FORMA DE RESILIENCIA ANTE L...
 
DRAC: Designing RISC-V-based Accelerators for next generation Computers
DRAC: Designing RISC-V-based Accelerators for next generation ComputersDRAC: Designing RISC-V-based Accelerators for next generation Computers
DRAC: Designing RISC-V-based Accelerators for next generation Computers
 
uElectronics ongoing activities at ESA
uElectronics ongoing activities at ESAuElectronics ongoing activities at ESA
uElectronics ongoing activities at ESA
 
Tendencias en el diseño de procesadores con arquitectura Arm
Tendencias en el diseño de procesadores con arquitectura ArmTendencias en el diseño de procesadores con arquitectura Arm
Tendencias en el diseño de procesadores con arquitectura Arm
 
Formalizing Mathematics in Lean
Formalizing Mathematics in LeanFormalizing Mathematics in Lean
Formalizing Mathematics in Lean
 
Introduction to Quantum Computing and Quantum Service Oriented Computing
Introduction to Quantum Computing and Quantum Service Oriented ComputingIntroduction to Quantum Computing and Quantum Service Oriented Computing
Introduction to Quantum Computing and Quantum Service Oriented Computing
 
Computer Design Concepts for Machine Learning
Computer Design Concepts for Machine LearningComputer Design Concepts for Machine Learning
Computer Design Concepts for Machine Learning
 
Inteligencia Artificial en la atención sanitaria del futuro
Inteligencia Artificial en la atención sanitaria del futuroInteligencia Artificial en la atención sanitaria del futuro
Inteligencia Artificial en la atención sanitaria del futuro
 
Design Automation Approaches for Real-Time Edge Computing for Science Applic...
 Design Automation Approaches for Real-Time Edge Computing for Science Applic... Design Automation Approaches for Real-Time Edge Computing for Science Applic...
Design Automation Approaches for Real-Time Edge Computing for Science Applic...
 
Estrategias de navegación para robótica móvil de campo: caso de estudio proye...
Estrategias de navegación para robótica móvil de campo: caso de estudio proye...Estrategias de navegación para robótica móvil de campo: caso de estudio proye...
Estrategias de navegación para robótica móvil de campo: caso de estudio proye...
 
Fault-tolerance Quantum computation and Quantum Error Correction
Fault-tolerance Quantum computation and Quantum Error CorrectionFault-tolerance Quantum computation and Quantum Error Correction
Fault-tolerance Quantum computation and Quantum Error Correction
 
Cómo construir un chatbot inteligente sin morir en el intento
Cómo construir un chatbot inteligente sin morir en el intentoCómo construir un chatbot inteligente sin morir en el intento
Cómo construir un chatbot inteligente sin morir en el intento
 
Automatic generation of hardware memory architectures for HPC
Automatic generation of hardware memory architectures for HPCAutomatic generation of hardware memory architectures for HPC
Automatic generation of hardware memory architectures for HPC
 
Type and proof structures for concurrency
Type and proof structures for concurrencyType and proof structures for concurrency
Type and proof structures for concurrency
 
Hardware/software security contracts: Principled foundations for building sec...
Hardware/software security contracts: Principled foundations for building sec...Hardware/software security contracts: Principled foundations for building sec...
Hardware/software security contracts: Principled foundations for building sec...
 
Jose carlossancho slidesLa seguridad en el desarrollo de software implementad...
Jose carlossancho slidesLa seguridad en el desarrollo de software implementad...Jose carlossancho slidesLa seguridad en el desarrollo de software implementad...
Jose carlossancho slidesLa seguridad en el desarrollo de software implementad...
 
Do you trust your artificial intelligence system?
Do you trust your artificial intelligence system?Do you trust your artificial intelligence system?
Do you trust your artificial intelligence system?
 
Redes neuronales y reinforcement learning. Aplicación en energía eólica.
Redes neuronales y reinforcement learning. Aplicación en energía eólica.Redes neuronales y reinforcement learning. Aplicación en energía eólica.
Redes neuronales y reinforcement learning. Aplicación en energía eólica.
 
Challenges and Opportunities for AI and Data analytics in Offshore wind
Challenges and Opportunities for AI and Data analytics in Offshore windChallenges and Opportunities for AI and Data analytics in Offshore wind
Challenges and Opportunities for AI and Data analytics in Offshore wind
 

Recently uploaded

Presentation on Food Delivery Systems
Presentation on Food Delivery SystemsPresentation on Food Delivery Systems
Presentation on Food Delivery Systems
Abdullah Al Noman
 
Asymmetrical Repulsion Magnet Motor Ratio 6-7.pdf
Asymmetrical Repulsion Magnet Motor Ratio 6-7.pdfAsymmetrical Repulsion Magnet Motor Ratio 6-7.pdf
Asymmetrical Repulsion Magnet Motor Ratio 6-7.pdf
felixwold
 
Prediction of Electrical Energy Efficiency Using Information on Consumer's Ac...
Prediction of Electrical Energy Efficiency Using Information on Consumer's Ac...Prediction of Electrical Energy Efficiency Using Information on Consumer's Ac...
Prediction of Electrical Energy Efficiency Using Information on Consumer's Ac...
PriyankaKilaniya
 
ITSM Integration with MuleSoft.pptx
ITSM  Integration with MuleSoft.pptxITSM  Integration with MuleSoft.pptx
ITSM Integration with MuleSoft.pptx
VANDANAMOHANGOUDA
 
Assistant Engineer (Chemical) Interview Questions.pdf
Assistant Engineer (Chemical) Interview Questions.pdfAssistant Engineer (Chemical) Interview Questions.pdf
Assistant Engineer (Chemical) Interview Questions.pdf
Seetal Daas
 
Beckhoff Programmable Logic Control Overview Presentation
Beckhoff Programmable Logic Control Overview PresentationBeckhoff Programmable Logic Control Overview Presentation
Beckhoff Programmable Logic Control Overview Presentation
VanTuDuong1
 
Levelised Cost of Hydrogen (LCOH) Calculator Manual
Levelised Cost of Hydrogen  (LCOH) Calculator ManualLevelised Cost of Hydrogen  (LCOH) Calculator Manual
Levelised Cost of Hydrogen (LCOH) Calculator Manual
Massimo Talia
 
Digital Image Processing Unit -2 Notes complete
Digital Image Processing Unit -2 Notes completeDigital Image Processing Unit -2 Notes complete
Digital Image Processing Unit -2 Notes complete
shubhamsaraswat8740
 
一比一原版(uoft毕业证书)加拿大多伦多大学毕业证如何办理
一比一原版(uoft毕业证书)加拿大多伦多大学毕业证如何办理一比一原版(uoft毕业证书)加拿大多伦多大学毕业证如何办理
一比一原版(uoft毕业证书)加拿大多伦多大学毕业证如何办理
sydezfe
 
Call For Paper -3rd International Conference on Artificial Intelligence Advan...
Call For Paper -3rd International Conference on Artificial Intelligence Advan...Call For Paper -3rd International Conference on Artificial Intelligence Advan...
Call For Paper -3rd International Conference on Artificial Intelligence Advan...
ijseajournal
 
Sri Guru Hargobind Ji - Bandi Chor Guru.pdf
Sri Guru Hargobind Ji - Bandi Chor Guru.pdfSri Guru Hargobind Ji - Bandi Chor Guru.pdf
Sri Guru Hargobind Ji - Bandi Chor Guru.pdf
Balvir Singh
 
OOPS_Lab_Manual - programs using C++ programming language
OOPS_Lab_Manual - programs using C++ programming languageOOPS_Lab_Manual - programs using C++ programming language
OOPS_Lab_Manual - programs using C++ programming language
PreethaV16
 
AN INTRODUCTION OF AI & SEARCHING TECHIQUES
AN INTRODUCTION OF AI & SEARCHING TECHIQUESAN INTRODUCTION OF AI & SEARCHING TECHIQUES
AN INTRODUCTION OF AI & SEARCHING TECHIQUES
drshikhapandey2022
 
Mechanical Engineering on AAI Summer Training Report-003.pdf
Mechanical Engineering on AAI Summer Training Report-003.pdfMechanical Engineering on AAI Summer Training Report-003.pdf
Mechanical Engineering on AAI Summer Training Report-003.pdf
21UME003TUSHARDEB
 
Impartiality as per ISO /IEC 17025:2017 Standard
Impartiality as per ISO /IEC 17025:2017 StandardImpartiality as per ISO /IEC 17025:2017 Standard
Impartiality as per ISO /IEC 17025:2017 Standard
MuhammadJazib15
 
Butterfly Valves Manufacturer (LBF Series).pdf
Butterfly Valves Manufacturer (LBF Series).pdfButterfly Valves Manufacturer (LBF Series).pdf
Butterfly Valves Manufacturer (LBF Series).pdf
Lubi Valves
 
Object Oriented Analysis and Design - OOAD
Object Oriented Analysis and Design - OOADObject Oriented Analysis and Design - OOAD
Object Oriented Analysis and Design - OOAD
PreethaV16
 
Supermarket Management System Project Report.pdf
Supermarket Management System Project Report.pdfSupermarket Management System Project Report.pdf
Supermarket Management System Project Report.pdf
Kamal Acharya
 
Ericsson LTE Throughput Troubleshooting Techniques.ppt
Ericsson LTE Throughput Troubleshooting Techniques.pptEricsson LTE Throughput Troubleshooting Techniques.ppt
Ericsson LTE Throughput Troubleshooting Techniques.ppt
wafawafa52
 
EV BMS WITH CHARGE MONITOR AND FIRE DETECTION.pptx
EV BMS WITH CHARGE MONITOR AND FIRE DETECTION.pptxEV BMS WITH CHARGE MONITOR AND FIRE DETECTION.pptx
EV BMS WITH CHARGE MONITOR AND FIRE DETECTION.pptx
nikshimanasa
 

Recently uploaded (20)

Presentation on Food Delivery Systems
Presentation on Food Delivery SystemsPresentation on Food Delivery Systems
Presentation on Food Delivery Systems
 
Asymmetrical Repulsion Magnet Motor Ratio 6-7.pdf
Asymmetrical Repulsion Magnet Motor Ratio 6-7.pdfAsymmetrical Repulsion Magnet Motor Ratio 6-7.pdf
Asymmetrical Repulsion Magnet Motor Ratio 6-7.pdf
 
Prediction of Electrical Energy Efficiency Using Information on Consumer's Ac...
Prediction of Electrical Energy Efficiency Using Information on Consumer's Ac...Prediction of Electrical Energy Efficiency Using Information on Consumer's Ac...
Prediction of Electrical Energy Efficiency Using Information on Consumer's Ac...
 
ITSM Integration with MuleSoft.pptx
ITSM  Integration with MuleSoft.pptxITSM  Integration with MuleSoft.pptx
ITSM Integration with MuleSoft.pptx
 
Assistant Engineer (Chemical) Interview Questions.pdf
Assistant Engineer (Chemical) Interview Questions.pdfAssistant Engineer (Chemical) Interview Questions.pdf
Assistant Engineer (Chemical) Interview Questions.pdf
 
Beckhoff Programmable Logic Control Overview Presentation
Beckhoff Programmable Logic Control Overview PresentationBeckhoff Programmable Logic Control Overview Presentation
Beckhoff Programmable Logic Control Overview Presentation
 
Levelised Cost of Hydrogen (LCOH) Calculator Manual
Levelised Cost of Hydrogen  (LCOH) Calculator ManualLevelised Cost of Hydrogen  (LCOH) Calculator Manual
Levelised Cost of Hydrogen (LCOH) Calculator Manual
 
Digital Image Processing Unit -2 Notes complete
Digital Image Processing Unit -2 Notes completeDigital Image Processing Unit -2 Notes complete
Digital Image Processing Unit -2 Notes complete
 
一比一原版(uoft毕业证书)加拿大多伦多大学毕业证如何办理
一比一原版(uoft毕业证书)加拿大多伦多大学毕业证如何办理一比一原版(uoft毕业证书)加拿大多伦多大学毕业证如何办理
一比一原版(uoft毕业证书)加拿大多伦多大学毕业证如何办理
 
Call For Paper -3rd International Conference on Artificial Intelligence Advan...
Call For Paper -3rd International Conference on Artificial Intelligence Advan...Call For Paper -3rd International Conference on Artificial Intelligence Advan...
Call For Paper -3rd International Conference on Artificial Intelligence Advan...
 
Sri Guru Hargobind Ji - Bandi Chor Guru.pdf
Sri Guru Hargobind Ji - Bandi Chor Guru.pdfSri Guru Hargobind Ji - Bandi Chor Guru.pdf
Sri Guru Hargobind Ji - Bandi Chor Guru.pdf
 
OOPS_Lab_Manual - programs using C++ programming language
OOPS_Lab_Manual - programs using C++ programming languageOOPS_Lab_Manual - programs using C++ programming language
OOPS_Lab_Manual - programs using C++ programming language
 
AN INTRODUCTION OF AI & SEARCHING TECHIQUES
AN INTRODUCTION OF AI & SEARCHING TECHIQUESAN INTRODUCTION OF AI & SEARCHING TECHIQUES
AN INTRODUCTION OF AI & SEARCHING TECHIQUES
 
Mechanical Engineering on AAI Summer Training Report-003.pdf
Mechanical Engineering on AAI Summer Training Report-003.pdfMechanical Engineering on AAI Summer Training Report-003.pdf
Mechanical Engineering on AAI Summer Training Report-003.pdf
 
Impartiality as per ISO /IEC 17025:2017 Standard
Impartiality as per ISO /IEC 17025:2017 StandardImpartiality as per ISO /IEC 17025:2017 Standard
Impartiality as per ISO /IEC 17025:2017 Standard
 
Butterfly Valves Manufacturer (LBF Series).pdf
Butterfly Valves Manufacturer (LBF Series).pdfButterfly Valves Manufacturer (LBF Series).pdf
Butterfly Valves Manufacturer (LBF Series).pdf
 
Object Oriented Analysis and Design - OOAD
Object Oriented Analysis and Design - OOADObject Oriented Analysis and Design - OOAD
Object Oriented Analysis and Design - OOAD
 
Supermarket Management System Project Report.pdf
Supermarket Management System Project Report.pdfSupermarket Management System Project Report.pdf
Supermarket Management System Project Report.pdf
 
Ericsson LTE Throughput Troubleshooting Techniques.ppt
Ericsson LTE Throughput Troubleshooting Techniques.pptEricsson LTE Throughput Troubleshooting Techniques.ppt
Ericsson LTE Throughput Troubleshooting Techniques.ppt
 
EV BMS WITH CHARGE MONITOR AND FIRE DETECTION.pptx
EV BMS WITH CHARGE MONITOR AND FIRE DETECTION.pptxEV BMS WITH CHARGE MONITOR AND FIRE DETECTION.pptx
EV BMS WITH CHARGE MONITOR AND FIRE DETECTION.pptx
 

Evolution and Trends in Edge AI Systems and Architectures for the Internet of Things Era

  • 1. Evolution and Trends in Edge AI Systems and Architectures for the Internet of Things Era Prof. David Atienza Alonso Embedded Systems Laboratory (ESL), EPFL david.atienza@epfl.ch FDI-Conferencia de Posgrado, UCM, 17 de noviembre, 2021 https://www.epfl.ch/labs/esl/
  • 2. © EPFL 2021  Team covers IoT and edge AI (both hardware and software co-design)  Headed by Prof. David Atienza  All started in 2008, 45 members today:  8 post-docs  22 PhD students (18 grad)  11 M.Sc. Theses (26 grad)  2 senior engineers  1 administrative assistant, 1 system admin.  Support from academia and industry for R&D projects  31 companies/R&D centers provide equipment, grants and donations David Atienza 2 ESL Team and Collaborations
  • 3. © EPFL 2021 Origin of IoT: Wireless Sensor Networks  Late 80s: Wireless Sensor Networks (WSNs) are data acquisition and distribution networks consisting of a set of (large) sensor nodes  Simple architectures (“get” data)  Cooperative applications and transmission David Atienza 1 2 Sink 3 4 T = 26.85 ºC T = 31.85 ºC T = 24.85 ºC T = 24.85 ºC is OK
  • 4. © EPFL 2021 In 2012 - The Internet of Things Concept : “Nano Sensors with Good Connectivity” David Atienza  Few more years beyond 1999, but thanks to Moore’s Law, after 45 years… Complexity increase Intel 4004 [1971] (2250 p-MOS trans.,108 KHz) Busicon Calculator 141-PF Intel 5th Gen Core i7 vPro [2012] (1.3B trans., 3.3GHz) 2012 1971
  • 5. © EPFL 2021 Wearables Connected Cars Connected City Industry 4.0 Transportation Healthcare Oil & Gas [Source: DARPA and EC] Continuous system monitoring (IoT sensors) David Atienza IoT System-Level Architecture Big-Data Cloud computing Connected Houses
  • 6. © EPFL 2021 David Atienza IoT Vision: Guardian Angels See more details at: https://www.ga-project.eu/
  • 7. © EPFL 2021 IoT Enabling Digital Era: Technology Convergence  IoT concept is the convergence of three technologies within “Digital Era” + Digitized Enterprise data 1 IoT sensors (Internet of Everything) Analytics and Deep Learning 3 Big data storage and Computing IoT platforms 2 Able to collect unprecedented amounts of data Capacity to store and handle large amounts of data Capacity to make sense of this data in a confident way David Atienza
  • 8. © EPFL 2021 IoT (Potential) Benefits David Atienza  IoT long-term economic benefits [McKinsey]  Remote healthcare: $11.1Trill/year saved (1B people)  Efficient energy: 45TWh/year saved in EU (4M homes)  Automotive industry supply chain: 50% costs reduction  … Business-to-business services: 70% added value! But… How can IoT be designed?
  • 9. © EPFL 2021  Two-fold paradigm shift in health delivery  Cardiovascular monitoring is key today… Environment Genetics Access to care Health behaviors, personal lifestyle Determinants of health issues (source: Institute for the future, Center for disease control and prevention, 2006) Symptom-based  Preventive healthcare Hospital-centered  Person-centered Trainer/coach Home record ECG Holter data logger (clinical practice) Resting Electrocardiogram (ECG) IoT to the Rescue of Our Healthcare System David Atienza
  • 10. © EPFL 2021 Apple Watch System: Example of (Expensive) IoT Sensor  April 2015: Combination of Watch + iPhone  IoT system design  Watch: “Raw” bio-signals acquisition  iPhone: User interface and long-range transmission  Cloud: Bio-signal (post-)processing David Atienza iPhone 6: A8 MPCore (Dual-core CPU, 4-core GPU) iPhone was less powerful! (ARM 1176JZF-S) Apple Watch: APL 0778 processor (with 4GB, 5 sensors, loudspeaker, …) Limited lifetime (< 5 hours for bio-signal analysis) and no real-time… COMMUNICATION OVERHEAD!
  • 11. © EPFL 2021 ECG Sensing and sampling Processing Communication Energy consumption breakdown  This ECG streaming monitor lasts 5h, why?  Send 16 bits: more energy than doing 64’000 operations! Long-Lived IoT Designs Feasible Today? David Atienza IoT age requires complete system revolution! How to reduce IoT data sensed and streamed? Only get and keep “useful” data from IoT sensors!
  • 12. © EPFL 2021 So, Is IoT Still on “Top of the Wave”?  Very high expectations since 2014… [Source: Gartner] David Atienza Did the “IoT dream” disappear?!
  • 13. © EPFL 2021 IoT just got “reborn”… [Source: Gartner] New IoT revolution, 3 technologies: (1) Edge AI (2) IoT platform (3) Deep learning (learn from Big Data!) David Atienza
  • 14. © EPFL 2021 New IoT Platform (“Smart Cloud”): 3 Delivery Modes  Infrastructure as a Service (IaaS): IT equipment  Platform as a Service (PaaS): Big Data access and DL libraries  Software as a Service (SaaS): Full AI application services Platform as a Service (PaaS) SaaS PaaS IaaS $ $$$$$ David Atienza
  • 15. © EPFL 2021 What is “Edge AI / Computing”? David Atienza 15  Intelligence is moving towards edge devices  Real-time pre-filtering and data processing  IoT devices send only a portion of the data to the central data repository, or corporate datacenter  Reliable results of Big Data AI on IoT Cloud platforms
  • 16. © EPFL 2021 David Atienza Does Data Size Matters for a Global IoT Business Perspective? 3x cost reduction if data volume is reduced by 95% Yes, only send (and keep) useful data: powerful IoT edge AI platforms
  • 17. © EPFL 2021 IoT Edge AI Concept: Complex Hardware Template David Atienza RF + COMPUTING
  • 18. © EPFL 2021 New Edge AI + Patches Can Understand Human Movements  New collaborative work with multiple patches and one IoT edge AI node  Multiple patches to track in real-time: but AI calibration needed David Atienza
  • 19. © EPFL 2021 Can you Use the edge AI Concept for new Businesses: Can you be a Ski Coach? Just Add An App and Earbuds!  Personalized coach with IoT sensors for ski  Personalized App  Create metrics based upon 4 fundamental skiing skills:  Edging,  Balance,  Rotation,  Pressure. David Atienza Based on few types of sensors and calibrated AI on the App, do you get something you can sell?
  • 20. © EPFL 2021 David Atienza See more details at: https://getcarv.com/ Can you Use the edge AI Concept for new Businesses: Can you be a Ski Coach? Just An App and Earbuds!
  • 21. © EPFL 2021 IoT Edge AI Sensing: Few Sensors for Many Application Domains  Sensors specific per application: but large convergence in similar products David Atienza
  • 22. © EPFL 2021  Highly-integrated Apple S6 SiP Chip  ARM DualCore SC300 @ 780 MHz  Integrated graphics and PowerVR (ML accel.)  32 GB of memory  Multiple sensors  GPS + GLONASS geolocat.  Heart Monitor sensor: PPG-based HR  Electrical heart sensor  3-axis accelerometer + gyroscope  Ambient light sensor, Built-in compass  Multiple radios included (+ eSIM)  Gold antenna embedded in casing  LTE/UMTS, NFC, Wi-Fi 802.11b/g/n, BT 5.0  Price: $399 Example (new) IoT Edge AI for Healthcare: Apple Watch Series 6 David Atienza New IoT Edge AI wearables (with limited set of sensors) provide minimal latency and more lifetime
  • 23. © EPFL 2021 What can new Edge AI Sensors for Healthcare do?  Same sensors, but algorithms and performance improved with edge AI accelerator chips David Atienza Accurate detection of cardiovascular pathologies without hospitals and specialized personnel!
  • 24. © EPFL 2021 Example IoT Edge AI for Smart Home Devices: “Personalized” Coffee  New IoT products already on the market… Control Edge AI, Big Data and DL IoT Edge AI saves energy and gives personalization: Is then connectivity still relevant? David Atienza Custom IoT Edge AI design: $3
  • 25. © EPFL 2021 New Open-Source Custom Hardware of ULP SoCs for IoT Edge AI Devices  Case Study: GAP8 Multi-Core SoC Edge Platform  RISC-V architecture (Based on ETHZ PULP) Source: GAP8 SDK internal documentation (https://github.com/GreenWaves-Technologies/gap_sdk) 1 general purpose core 8-core cluster for complex signal processing A lot of memory 28nm FD-SOI New Deep Learning accelerators Custom IoT Edge AI design: $3
  • 26. © EPFL 2021 Vertical Companies Created to Control New IoT Space  IoT business Landscape: Data-related services more profitable David Atienza Source: Yole Dev. ’17 (Upd.)
  • 27. © EPFL 2021  IoT business Landscape: Data-related services more profitable  But largest companies trying to avoid risks: full chain control David Atienza Vertical Companies Created to Control New IoT Space Amazon Echo Amazon Zero Touch
  • 28. © EPFL 2021 New Extended Support in Cloud for Secure IoT Devices: Example of Amazon Web Services (AWS IoT) and Google Teachable Machines  Open-source IoT communication standards supported: Message Queuing Telemetry Transport (MQTT)  Enable new IoT persistence concept: Device Shadow concept (Virtual Twin) when disconnected (Official) Secured devices David Atienza
  • 29. © EPFL 2021 New IoT Revolution… Enabling Industry 4.0! … Collect data from everywhere … Exploit all sources to make decisions … New ways to interact with machines … New robotics help deliver efficiency Monitoring (all) and maintenance Virtual twins and model predictions Human-Machine interfaces Digital autonomy in physical world Edge IoT Big Data analytics Wearables Augmented reality Advanced robotics Deep learning Deep learning Edge IoT Edge IoT Deep learning Edge IoT David Atienza
  • 30. © EPFL 2021 New Robotics Systems for IoT: Edge AI + Sensing Evolution for Real-Time Reaction  From frame- to event-based Dynamic Vision System (DVS): analog frontend off most of the time Frame- based {x 0 , y 0 } {x 1 , y 1 } {x 2 , y 2 } {x 3 , y 3 } {x N-1 , y N-1 } Event- based David Atienza PULPv3 GrainCam Mixed-Signal Event-based Imager Digital Parallel Processor [Courtesy: L. Benini]
  • 31. © EPFL 2021  Insightness: Obstacle avoidance with 7-layer Deep neural network  Learn with public traffic databases Delivery with Drones: React in Real-Time with IoT Edge + Deep Learning [Courtesy: Davide Scaramuzza, RPL-Univ. Zurich] David Atienza
  • 32. © EPFL 2021 IoT just got “reborn”… But keeps evolving in 2021! David Atienza [Source: Gartner] But 3 techs. of IoT revolution remain: (1) Edge AI: Embedded AI, carbon-based transistors, biodegradable sensors, ... (2) IoT platforms: SASE, data fabric, digital twin of the person, citizen twin, composable infrastructure or enterprise, ... (3) Deep learning/AI: Explainable AI, Responsible AI, Generative AI, Composite AI, Adaptive ML, GANs,.. (4) Communication key again: 5G?
  • 33. © EPFL 2021 New 5G Connectivity for IoT: Between Local vs Global Networks Today: Local Networks (Legacy)  Simple local networks, short range links (scalable so far, but with latency)  Requires local infrastructure devices, setup, and maintenance  Transceiver simple/cheap  No extra service subscription (low rates)  Reuse of existing comm. resources Global Connectivity (5G)  Complex (cellular) network with long range links  No local infrastructure with some QoS by service provider  Complex/costly transceiver  Requires service subscription  Limited reuse, no shared comm. resources David Atienza Evolution promised/envisioned by 5G [Courtesy: Andreas Burg (TCL@EPFL)]
  • 34. © EPFL 2021 Conclusion  New IoT concept possible due to techs. convergence: real-time response, safer and better AI  Upcoming technology trends:  Even more distributed/hierarchical intelligence in future IoT Era  “Copy” more from biology (More specialization on processors = energy efficiency, etc.),  Finding schemes to secure and activate selectively IoT devices (Self-awareness) 1 Edge AI sensors (and actuators) Analytics and Deep Learning 3 Big data storage and Computing IoT platforms 2 Unprecedented amounts of (useful) data at the edge/fog Capacity to store and handle large data volumes Capacity to make sense of it (automatically) David Atienza