Jan M' Rabaey and the PicoRadio Group

1

• Jan M. Rabaey and the PicoRadio Group
• EECS Dept.
• Univ. of California, Berkeley

2
PicoRadio Team

• Faculty
• Jan Rabaey
• Paul Wright
• Kannan Ramchandran
• Ali Niknejad
• Alberto Sangiovanni-Vincentelli
• Adam Wolisz (UT Berlin)
• Testbed
• Fred Burghardt
• Sue Mellers
• Jonathan Reason
• RF
• Yuen Hu Chee
• Brian Otis
• Nathan Pletcher
• Pavel Monat
• Energy Scavening and Packaging
• Mike Montero
• Dan Odell

• Design Tools
• YanMei Li
• Rong Zhong
• Networking, MAC and Locationing
• Delynn Bettencourt
• Alvise Bonivento
• Tufan Karalar
• Enyi Lin
• Dragan Petrovic
• Rahul Shah
• Jana Van Greunen
• Charlie Zhong
• Network Processor
• Josie Ammer
• Huifang Qin
• Mike Sheets
• Jonathan Tsao
• Allan Tsao
• Ruth Wang

collaborative effort started with NEST
3
PicoRadios - The Original Mission

• Meso-scale low-cost radios for ubiquitous
  wireless data acquisition that
• are fully integrated
• Size smaller than 1 cm3
• minimize power/energy dissipation
• Limiting power dissipation to 100 mW
  enables energy scavenging
• and form self-configuring ad-hoc networks
  containing 100s to 1000s of nodes

Pushing the limits ever further!
Main targets ambient intelligence, smart homes
4
The Quark Node (expected late summer)
5
Optimizing the supply network
6
Where are we heading?The PicoRadio Program in
Short
Energy, cost and size awareness at all the levels
of the abstraction chain
Towards a universal, scaleable and portable
application interface for AWSANs
Testbeds
Applications and Applications Support
Networking, Applications, Media Access, and
Positioning
Unified energy and power efficient protocol stacks
Picoradio Computational Platforms
Bringing power/energy levels down to
unprecedented levels
PicoRadio RF and Baseband
Energy Generation
Testing and prototyping all of the above in
realistic environments
7
A Service-based Universal Application Interface
Application Space
Application Instance
Platform Mapping
AI Platform
Application Interface (AI)
Platform Design Export
Platform Instance
Architecture Space

• Universal independent on the Implementation on
  any present and future Sensor Network Platform
• Service-based standard set of Services and
  Interface Primitives available to Applications
• Analogy with Internet Sockets

Proposal submitted to NSF (with A. Wolisz, K.
Ramchandran, M. Franklin, A. Sangiovanni)
8
The AWSAN as a Distributed Database
The query as the basic access mechanism
(TinyDB)Get the temperature in the kitchen
QS allows a controller to obtain the state of a
group of components
9
Using Semantic Addressing

• Name definition
• attribute specification attribute selector
  expression
• e.g. temperature, gt 25, humidity, gt75 R.H.,
  temperature gt 25 OR humidity gt 75 R.H.
• scope
• Region (e.g. kitchen, BWRC, Berkeley)
• Organization (e.g. PGE)

• Names are not unique
• Names may change during network operation

Enables ad-hoc operation and provides robustness
10
Prototype The Demand-Response Application
Distributed Control
Sensors
One or more controllers (glorified thermostats)
Actuators
Energy sensors Temp Sensors
Determine what and when to turn on/off based on
readings from sensors and do this using actuators
CALISO
WWW
Virtual sensors
Virtual actuators
11
Networking, Application, Media Access, and
Positioning

• Goal Providing an energy efficient, truly ad-hoc
  and absolutely reliable network stack from
  physical to application layer
• Our (not so) secret weapons
• Embracing randomness and exploiting redundancy,
  avoiding anything that leads to brittleness (such
  as distributed state)
• Networking strategies that are energy-aware and
  networks that dynamically adapt their properties
  based on energy availability
• Media-access protocols to maximize sleep-time,
  minimize monitoring and synchronization overhead
• A group of supporting services (such as time
  synchronization and positioning)
• A supporting design methodology (simulation,
  modeling, emulation)

3 papers at ICC in Paris (June 04)
New TELOS motes (Culler) the selected platform
for us to experiment with some of these ideas
12
Combining Energy-Efficiency and Reliability
Through Redundancy and Randomness
Developed complete modeling and simulation
environment (Omnet, Matlab, BEE)
13
Telos A Platform for Prototyping
Developed by D. Culler group (NEST)

• Standards Based
• IEEE 802.15.4
• USB
• IEEE 802.15.4
• CC2420 radio
• 250kbps
• 2.4GHz ISM band
• TI MSP430
• Ultra low power
• 1.6mA sleep
• 460mA active
• 1.8V operation
• TinyOS support

• A new platform for low power research
• Monitoring applications
• Environmental
• Building
• Tracking
• Long lifetime, low power, low cost, integrated
  sensors
• Adavnatages over current Mica platform?
• Oscillator start up times, oscillator noise, and
  operating/sleep current too high
• Instead leverage 802.15.4 low power operation and
  newMCUs

Courtesy J. Polastre
Will be used for real-world prototypes of SNSP
and Protocol stacks Plan to integrate new
Picoradios into the platform
14
Auxiliary Services provide Sense of Space, Time
and Concept Essential for True Ambient
Intelligence Experience
15
Ultra-Low Power Design

• Towards a sub-100 mW integrated node

Power Supply Network
Baseband (mixed-signal)
ClockGeneration
RF Antenna
DigitalProcessor
16
Operational Super-Regenerative Receiver

• Fully Integrated no crystal or off-chip
  inductors
• Prxlt400mW
• Sensitivity -100dBm

2mm
1mm
fq100kHz
Super-
OOK
Non-Linear
PWM
Isolation
Regenerative
digital chip
Detector
Demodulator
Amplifier
Filter
Oscillator
Vinfc1.9GHzfq100kHz
Iinfc1.9GHz
Vinfcfq100kHzPWM signal
Vout10kbpsOOK signal
Pinfc1.9GHz10kbps - OOK
17
Results

• Prx lt 400 mW
• -3dB BW1MHz
• Sensitivity -100dBm (12dB SNR)
• Intrinsic AGC
• S11 -10dB _at_ 1.9GHz
• Radiation -60dBm

-80 dbM OOK
18
Energy-efficient Transmitters

• Power Oscillator to deliver power efficiently and
  reduce driver power (self-driven)
• FBAR Reference Oscillator
• Concurrent Antenna/Power Oscillator design to
  provide optimal load termination
• Power Control for optimal radiated power
• Frequency Calibration to minimize locking power

Injection-locked transmitter
Core Devices
7-bits capacitive array
TX at 2 mW or less (when on)
19
ULV Oscillators for Reactive Radios
Sub-threshold RF oscillatorusing integrated LCs
Lower parasitics
Lower parasitics
2 different inductors
Higher Q
2.4 nsec startup
Should enable Sub-100 mW receivers
20
Baseband Processing

• Mostly Digital

Preliminary Data!
Mostly Analog
with Fernando De Bernardinis, Yanmei Li, and
Brian Otis
21
Ultra Low-Voltage Digital Design
Goal Operate next generation PicoNode at 300 mV
or below
Energy-performance trade-off

• Challenges
• Maintaining performance
• Dealing with process variations
• Keeping the memories operational (data retention)

Adaptive tuning of Vdd and Vt essential Use
On-chip tester!
Yoda group Huifang Qin, Ruth Wang, Paul
Friedman, Andrei Vladimirescu, (Costas Spanos,
Tsue-Jae King) (Potential project with DARPA MTO)
22
Some aggressive solutions
And many other others. See http//bwrc.eecs.berke
ley.edu/Research/yoda/
23
The Memory Challenge
The Data Retention Voltage
Spatial distribution of DRV

• Solutions
• Sizing
• Advanced error correction
• Other memory architectures
• Should bring us below 150 mV

Best paper award at ISQED 2004
24
Reducing the Clock Power
Co-design of an integrated CMOS/MEMS 16MHz
reference oscillator (in collaboration with E.
Quevy BSAC)
VP3.3V Lx 0.2015H Cx0.4822fF
Rx2.044kW Rp1MW Q10,000
Electrostatic Gap 50nm Process Sacrificial Ge
Blade process
Received Best Technology Price at Recent Haas
Business Plan Competition
25
Reducing The Clock Power
MEMS resonator die flips directly onto CMOS for
low interconnect parasitics and a compact,
integrated clock module.
Resonator outline
Series Osc
Parallel Osc
Ultra-low power (1 mW)
Low-phase noise
26
Powering Ultra-Dense Networks
Needs integrated meso-scale energy train
Micro-battery
Electrostatic MEMS vibration converters
Proposal submitted to NSF (PIs Rabaey, Wright,
Sanders, Steingard, Roundy)
27
Perspectives

• Energy-scavenging wireless sensor and actuation
  nodes are within reach
• - Break-out What to do with a mW?
• Use of aggressive and innovative manufacturing
  techniques can lead to sub-1 solutions
• Break-out what to do with 1 million nodes?
• PicoRadio technologies are ready for prime time!
• The main challenge building purely ad-hoc
  reliable networks that are portable
• Towards the future another factor 10 in size and
  power
• Leading to complete new paradigms in computation
  and computation (See D. Pretrovic presentation
  this afternoon)