Transcript Slide 1

Smart Dust: Unique Low Power
Flexible Sensor Network
Neil Goldsman, Haralabos (Babis) Papadopoulos
and Shuvra Bhattacharyya
Dept. of Electrical and Computer Engineering
University of Maryland
College Park
Lead Personnel
All UMD ECE Faculty
•Prof. Neil Goldsman
–Director of ECE Mixed Signal VLSI Lab
–Director of ECE Semiconductor Simulation Lab
–Introduced RF VLSI Circuit Design to UMD
–Other Areas: 3D Integration
•Prof. Haralabos (Babis) Papadopoulos
–Institute for Systems Research
–Communications and Signal Processing Laboratory
–Center for Satellite Communications
–Other Areas: Power Efficient Private Comm. Networks
•Prof. Shuvra Bhattacharyya
–Institute for Advanced Computer Studies
–Digital VLSI Design Automation Laboratory
–Embedded Systems Research Laboratory,
–Communications and Signal Processing Laboratory
–Other Areas: Embedded Software; Low Power Design
Outline
•
•
Smart Dust Overview
Electronics (Neil)
–
•
Communications (Babis)
–
•
RF VLSI, Hardware, Sensors and 3D Integration
Algorithms and Signal Processing
Microcontrol (Shuvra)
–
Digital Hardware and Software for Data Flow and
Processing
Overview: Smart Dust Network
•A network of smart sensors (dust
particles) that communicate with each
other wirelessly and perform distributed
computations.
•Dust particle to be mm size (grain of
sand).
•Network to be seamlessly integrated
into environment for flexible application.
•Each dust particle usually contains a
sensor, a micro-controller, a transceiver,
and powering mechanisms
•The network can contain several hundreds
or even thousands of dust particles.
Overview: What is a Smart Dust Network?
• The dust particles sense physical attributes of
the environment, and then wirelessly
communicate these attributes to other dust
particles.
• Using a distributed computing algorithms, the
dust particles fuse their data, and make
decisions based on the cumulative information
of the overall network.
• Applications:
–
–
–
–
Motion and Distance tracking
Biological and Chemical Environmental Factors
Distributed Image Recognition and Optical Sensing
Acoustic and Vibrational Sensing
Overview: The Challenges
• Dust particles are small size but must still house RF
and computational circuitry.
• Low Power: Dust particles are either to be powered
externally or by a power source that is
commensurately sized with respect to the dust
particle. We must perform RF communication,
computations and sensing which minimize power
use, but still overcome noise limitations and bit
errors.
• How to physically produce dust particles that allow
for sensing, communication and computation on a
system that is millimeter scale?
• What special computational codes and hardware do
we need, and what special communication algorithms
do we need?
Overview: Overcoming the Challenges
To develop a state-of-the-art smart dust system
we have divided the project into three major
tasks.
1. Design of Physical Smart Dust Particles
(Neil).
2. Development of Network Communication
Protocols and Fusion of Data (Babis).
3. Computation for Data Analysis and Network
control (Shuvra).
Design of Analog Smart Dust Particle
Components
Neil Goldsman
Dust Particle Basic Topology
Achieve with State of Art CMOS, Novel Materials (CNTs) &
Unique 3D Integration
L1
Sensor
ADC
μP
L2
•Sensor: Optical, Biological, Temperature, etc. (analog)
•μP: microprocessor (control and signal processing)
•Tx, Rx: Transceiver
Tx
Rx
Use Phase-Locked Loop-Based Transceiver
Rx
BPF
DEMOD
LNA
IMF
IFF
1,0
Series
to Parallel
Sample CLK
f1-f2=
f2-f3=
f3-f1=
….
PLL
N,N+1,N+2,…
N,N+1
Tx
PA
MOD & Carrier
CPU
Background: FSK PLL VLSI Chip Already
Developed
PLL
Reference
Frequency
Dual Modulus
Prescaler
Digital
Input
Analog
Input
A/D
Converter
Shift
Register
Clock
Data
Sampling
Rate
Output
Output
Driver
Output
Driver
VCO
PFD
12-Bit Counter
Background PLL FSK Chip Prototype
Fabricated in 0.5μ CMOS
VCO
VCO
Output
Driver
Digital Switching Noise Testing Circuit 1
PLL Output Measurements
Low Phase Noise, Nice Lock
Frequency:
Phase
Noise:
50MHz
100MHz
500MHz
-1.6dBc/Hz
-1.8dBc/Hz
-2.2dBc/Hz
20.8dBc/Hz
16.5dBc/Hz
-1.4dBc/Hz
Analog Structures for Meeting Low Power,
Small Size Dust Design Constraints
• Operating frequency high GHz for small dimensions
• On-Chip antenna tuned to high GHz
• Transceivers rely heavily on tuned circuits
– Need State of Art Inductors for Max Q and Min Area.
– 3D inductor and Antenna design
– Carbon Nanotube Kinetic Inductor (2nd generation)
• Sensors
– On chip temperature sensor array
– Optical diode array (visible and infrared)
– MEMS accelerometer, microphone (2nd generation)
• On-chip microwave power harvester
– Antenna and Schottky diode form power supply
Assemble into Single Dust Particle with
Novel 3D Integration Methods
Achieving Smart Dust with 3D Integration
• To integrate sensors, RF,
and processors into a
millimeter size system,
3D integration is a likely
approach.
• We can build on our
existing research in this
area with LPS (G. Metze).
• Integration achieved
through modular levels.
Three-Dimensional
Integrated Dust Particle
3D Inductor and Transformer Test Structures
Prototype fabricated in 0.5μ CMOS
with MOSIS.
Transformer
Simulations indicate 3D structures
have much larger L and Q
Currently being verified
experimentally a with network
analyzer.
Planer
Spiral
Inductor
EMC
between
Spiral and
Transistor
3D
Spiral
EMC between
Transistors
EMC between
n wells
Prototyping Temperature Sensor Array (10x10)
Diodes are used as temperature
sensors:
Diode current increases
exponentially with temperature.
Fabricated in 0.5μ CMOS
3-D Prototyping Stacking Chip
•New chip with
structures that can
connect in 3D.
•Uses Symmetric Vias
for 3D Interconnects”
•Prototype
fabricated in 0.5μ
CMOS with MOSIS.
Proof of Principle Prototype
• Develop Temperature Sensor Smart Dust
Network
• Use aforementioned RF and Temperature
Systems
• Employ State of the Art Dust with 3D Integration
• Power
– First with batteries
– Second by harvesting the environment (RF, Heat)
• Communication and Microcontrolling schemes to
be used are described next.