Smart Dust - PowerPoint PPT Presentation

View by Category
About This Presentation
Title:

Smart Dust

Description:

In a static system, identification could be saved in mote memory ... Use specific motes as lookup servers for mapping the network. Disseminate lookup ... – PowerPoint PPT presentation

Number of Views:8242
Avg rating:3.0/5.0
Date added: 3 July 2020
Slides: 38
Provided by: DNS
Learn more at: http://www.cs.huji.ac.il
Category:
Tags: dust | mote | smart

less

Write a Comment
User Comments (0)
Transcript and Presenter's Notes

Title: Smart Dust


1
Smart Dust
Embedded Computing Seminar
Noam Sapiens
2
Outline
  • What is smart dust?
  • Characteristics
  • Applications
  • Military
  • Commercial
  • Requirements and restrictions
  • Analysis of smart dust communication
  • General architecture and design
  • What we have today
  • Would like to have
  • References

3
What is Smart Dust?
Large scale networks of wireless sensors for
various applications
  • The three key capabilities of smart dust are
  • Sensory capabilities
  • Processing capabilities
  • Communication capabilities

4
Smart dust characteristics
  • A system is made of one or a few base stations
    (interrogators) and as many smart dust motes as
    possible or required
  • Ubiquitous sensors of different types
  • Very task/application oriented design and
    performance
  • Wireless communication
  • Self-organizing, self-optimizing,
    self-configuring, self-sustaining.
  • Very small (should be under 1mm3)
  • Low power consumption
  • Easy to deploy
  • Based on current or very near future components

5
Military and Space applications
  • Internal and external spacecraft monitoring
  • Meteorological and seismological monitoring in
    difficult terrain and environments
  • Land/space communication
  • Chemical/biological environment sensing
  • Meteorological sensing for better aiming of
    guns and artillery
  • Autonomous vehicles external aid

6
  • Surveillance
  • Sensors minefield e.g. smart clear tracks on
    borders
  • Urban engagement (cont. DARPA funding in 2005)
  • Motion detection and enemy numbers
  • Bunker/building mapping
  • Peace time/treaty monitoring
  • Intelligence in hostile areas/behind enemy lines
  • Transportation monitoring and traffic mapping
  • Missile hunting
  • Monitoring soldier vitals and injury
  • Pursuit aid

7
Unmanned pursuit
Integration of several smart dust experiments
  • Aerial smart dust deployment in the area of
    interest ground and air
  • Sensors
  • Each mote has motion detectors and a small CMOS
    camera
  • Some motes has GPS
  • Computation
  • Image processing for target distinction
  • Communication
  • Ad-hoc networking
  • Relative localization

Energy tradeoff
Local coordinate system
Northwestern university
UC Berkeley and MLB Co.
8
UC Berkley PEG (pursuit-evasion game) experiment
  • 200 sensors network
  • One aerial and three ground unmanned vehicles
    pursuers
  • One ground unmanned evader
  • Pursuers are interrogators of the sensor network
    deployed
  • Sensor networks roles
  • Provide complete monitoring of the environment,
    overcoming the limited sensing range of on board
    sensors
  • Relay secure information to the pursuers to
    design and implement an optimal pursue strategy
  • Provide guidance to pursuers, when GPS or other
    navigation sensors may fail

UC Berkeley
9
Experiment block diagram
Sensor Network
EvaderDynamics
Pursuer
PursuerDynamics
GPS
Evader motionestimator
Pursuit Strategy
Tracking control
10
Methodologies
  • Autonomous sensor nodes deployed
  • Target vehicles traverse sensor field
  • Clumps of sensors exchange information
  • Data association from local information
  • Clump estimates target heading, speed, position
  • Computations use robust Closest Point of Approach
    statistics
  • Target parameters used to match existing tracks
  • Euclidean metric finds track with best-fit
  • New parameters merged with existing ones
  • Track information reported to user workstation
  • Track information propagated in advance of target
  • Diffusion routing limits information propagation
  • Difficult global problem decomposed into
    tractable local problems

11
Tracking process demonstration
Track initiated and users told
Target moves and detected
Target detected
Nodes exchange readings
Clump head selected
Track info propagated
Readings exchanged
Clump head chosen
Track updated and user told
Track info propagated
Recourse
Penn State University and DARPA
12
Experimental results
Unsuccessful tracking
Successful tracking
Penn State University and DARPA
13
Commercial applications
  • Games and sports
  • Traffic monitoring
  • Inventory control
  • Security
  • Identification and tagging
  • Predictive maintenance
  • Product quality control
  • Industrial facilities
  • Vehicles and systems
  • Appliances
  • Agriculture

14
  • Building management
  • Energy management
  • Temperature control
  • Lighting control
  • Fire systems
  • Smart office spaces
  • Computer interface
  • Virtual keyboard
  • 3D virtual sculpturing
  • Health, medicine and wellness
  • Handicap aid

15
(No Transcript)
16
Requirements
  • Perform a specific task according to the
    application
  • Sense as defined by the task profile (different
    types of detectors will not be discussed in
    this talk)
  • Perform basic computations digitization, noise
    filtering, DSP, FFT, image processing, decision
    making, localization, etc
  • Establish ad-hoc communication in a physical
    environment
  • Base station communication and peer to peer
  • Ranges between a few meters (between motes) and
    over a km (motes to base station)
  • Multi-hop routing (if required)
  • Self configuration and optimization

17
Restrictions
  • Mote volume will not exceed 1mm3
  • A single mote is probably restricted to few
    sensory capabilities
  • Energy restrictions
  • Battery 1J/mm3 (about 10?W for a day)
  • Capacitors 1mJ/mm3
  • Solar cells 1J/day (sun) or 1mJ/day (room
    light)
  • Vibrations 0.4-30?W (depends on amplitude and
    frequency)
  • Thermopile 0.4-2?W _at_ 25-37?C
  • Very low cost motes (enable large scale
    distribution)
  • No science fiction technologies

18
Some basic energy data
  • Digital calculations (e.g. writing/reading
    to/from memory, magnetic (memory) or electronic
    (transistors and gates) manipulations, Boolean,
    arithmetic etc.) 1pJ/bit
  • Analog circuitry (e.g. amplification) 1nJ/amp
  • DAQ 1nJ/sample (or passive in some sensors)
  • A/D and D/A 1nJ/instruction
  • MEMs control 1pJ/bit _at_ 1kb/sec

19
Analysis of smart dust communication
RF vs. Optical
  • RF radio frequency
  • MHz hundreds of GHz ? 1mm 100s meters
    wavelength
  • Technologies
  • Bluetooth
  • Cell phones (GSM, CDMA, etc.)
  • RFID
  • Optical
  • 100THz 1PHz ? 0.3? - 1.6? wavelength
  • Lasers and LEDs

20
RF
  • Pros
  • Well developed technologies
  • Multiplexing techniques TDMA, FDMA, CDMA.
  • Does not require line of sight
  • Not much affected by the environment
  • Cons
  • Antenna size (has to be at least ¼ of the
    wavelength)
  • Complex circuitry (modulation/demodulation,
    bandpass filters, etc.)
  • Energy consumption (approx. 100nJ/bit)

21
Optical
  • Pros
  • Low energy consumption (lt1nJ/bit)
  • High data rates
  • Small aperture, very directional (localization)
  • Spatial division multiplexing
  • Cons
  • Very directional
  • Line of sight
  • Atmospheric turbulence, weather and
    environmental conditions dependent

22
General smart dust mote architecture - optical
23
MEMs controlled corner cube retro-reflector
  • Perfectly aligned corner cube reflects light at
    the exact same direction of incidence
  • MEMs control of one of the corner cube sides
    alignment enables modulation
  • Energy consumption of about 1pJ/bit _at_ 1kb/sec
  • Range up to 1km

UC Berkeley
24
Smart dust active transmitter
  • Incorporates a laser, lens and a MEM steering
    mirror
  • 1mrad transmission
  • Data rate of approx. 5Mb/sec
  • Energy consumption depends on distance and
    detector size

Distance Detector area Energy consumption
5m 0.1mm2 20pJ/bit
5km 1cm2 10nJ/bit
500km 1m2 25nJ/bit
1mW at 1mrad laser is 40 times brighter than 100W
light bulb
25
SEM view
Laser diode
Lens
MEM mirror
Optical view
UC Berkeley
26
The different parts
The laser diode
The MEM mirror
The micro-lens
27
Experimental results
  • Beam steering at kHz rates
  • Steering in approx 1str 60?X 60?

5.2 km Berkeley Marina
15.3 km Coit Tower
300m Link test
14?W laser
8mW laser
28
The base station
  • Hand held
  • Binoculars
  • Palm
  • Cell phone
  • Laptop computer
  • Command center
  • Unmanned vehicle (land, sea, air)
  • Autonomous systems

29
Base station architecture
Quarter-wave
Filter
Polarizing
Plate
Beam Splitter
Lens
Camera
Smart dust
Beam
Laser
Expander
Mirror
Optical interrogation principles of operation
For exampleFOV17mX17mCMOS is 256X256, 43?2
pixelsRange 2kmfLens20cmSpatial resolution
6.6cm2
Space division multiplexing
30
Airborne base station example
UC Berkeley and MLB Co.
31
Challenges for mobile networking for smart dust
  • Line of sight requirement
  • Link directionality
  • Parallel readout and cross talk
  • Trade-offs
  • Revisit rates

32
Line of sight requirement
  • Optical communication requires photons from the
    transmitter reach the receiver photons travel
    in straight lines
  • Line of sight is not the only way of making the
    photons arrive at a desired location
  • Diffuse reflections low energy, wide spread
    (the entire FOV) and low contrast with the
    environment (especially with interrogating beam)
  • Non fixed smart dust systems - line of sight
    could be achieved intermittently
  • Ad hoc multi-hop routing

Cannot work with passive communication, very
small SNR
Latency
Algorithms Latency Reliability
33
Link directionality
General
  • Motes are unaware of neighbors location
  • Base station can disseminate location
    information to motes

Passive links
  • A corner cube retro-reflector angle of
    acceptance is 10-20?
  • Placing multiple corner cubes
  • Placing the corner cube and the receiver on a
    MEM mount signal maximization
  • Increase mote density high probability for
    communication with at least some motes in the
    area of interest

34
Active links
  • Mote receiver is omnidirectional within a
    hemisphere
  • Enables mote attention without aiming
  • No source identification
  • Making the receiver directional (by adding a
    lens) and connecting its directionality to the
    transmitter will enable communication
    automatically to the source
  • Requires aiming
  • Solved by increasing the density of motes
  • In a static system, identification could be
    saved in mote memory
  • Difference between receiver and transmitter
    angular spreads leads to non-reciprocal linking

35
Formation of smart dust self-organizing networks
User
GUI
Self-Organizing Sensor Application Systems
Database Server
Sensor Data Repository Manager
Distributed Sensor Query Processing
Collaborative Signal Processing
  • Remote service execution
  • Event notifications
  • Mote service discovery
  • Change detection scheme
  • Trigger management
  • User-defined adaptation
  • handler
  • Group management
  • Dataflow and group
  • structure
  • Group communication
  • Group reconfiguration

Configurable Distributed Services
Distributed Lookup Service
Distributed Composition Service
Distributed Adaptation Service
Dynamic Sensor Network
Event-based Diffusion Network
DARPA/ITO
36
Diffusion networks
  • Assume self awareness
  • Scan for neighbors according to criteria and
    scanning algorithms
  • Notify neighbors of your existence
  • Notify previously know neighbors about new
    neighbors found
  • Resembles click a link internet browsing

37
Lookup services
  • Use specific motes as lookup servers for mapping
    the network
  • Disseminate lookup information to relevant motes
  • Use region filters to reduce network traffic and
    avoid irrelevant connections

Experimental results
Average Response Delay
Average Network Traffic
Client Average Throughput
Resembles for example Google crawlers
DARPA/ITO
38
Composition Services
  • Use specific motes as composition servers
  • Create groups dynamically for local
    collaboration
  • Maintain group communication
  • Connection between tasks
  • Data flow monitoring and control (split, merge,
    filter, buffer)

DARPA/ITO
39
Adaptation Services
DARPA/ITO
  • Application renders a condition trigger, and
    adaptation handler changes network algorithms
  • Dynamic steering Distributed sensor
    applications steer around changes in the sensor
    network, such as mobility, failure, density,
    certainty, and reconfiguration
  • Dynamic clustering Active re-clustering of
    sensors based on density and level of activities
    to reduce collaborative processing and
    communication costs
  • Dynamic tasking Implement changes in task
    requirements of fielded sensors by dynamically
    downloading and executing codes to targeted
    sensors

40
Parallel readout and crosstalk
  • The network architecture of smart dust enables
    space division multiplexing in the base station
  • There are as many channels as there are pixels
    in the CMOS camera of the base station
  • If the interrogating beam is divergent enough
    several motes could be ready simultaneously
  • A base station will not distinguish between
    motes in the same space equivalent pixel
  • TDMA could be incorporated in the architecture
    modulation of the interrogating beam could
    establish a clock for synchronization
  • Demand access method (as in cellular and
    satellite networks) could be implemented as well
    a mote sends an active short pulse to the base
    station will receive attention by the
    interrogation beam of the base station

41
Trade-offs
SNR signal to noise ratio, governs the
probability for bit error Pt average
transmitter power A receiver area N0 receiver
inherent noise B bit rate r the distance
between the transmitter and receiver ? - beam
divergence
42
Revisit rate
  • Revisit rate should be application specific
  • Use of AI learning system
  • Frequent revisits to areas in which changes
    happen most rapidly
  • Could be based on human judgment or automatic
  • Could be based on the demand access method

43
What we have today
www.dust-inc.com
www.xbow.com
  • Different markets
  • Airborne systems monitoring, camera stability,
    unmanned
  • Marine
  • Land vehicles
  • Environment
  • Mote price 100
  • Kit price (8-12 motes) 2000
  • Building management
  • Industrial monitoring
  • Security

44
Smart dust assembly no science fiction
technologies
Full clip
UC Berkeley
45
A different type of smart dust is it really?
Chemical sensor active smart dust based on
material engineering
Chemical sensors
Chemical containment
Water drops manipulation
UC San Diego
46
Would like to have capabilities (a partial list)
  • Miniaturization of available smart dust and
    extreme price reduction
  • Possibility of optical pre-processing and
    optical circuits
  • Incorporate the concept of smart dust societies
    integration of different types of smart dust
  • Requires more robust network protocols
  • Requires better definition of mote task
  • Enables complex systems easy distribution
  • Enables smaller and cheaper motes

47
  • Multi wavelength VCSEL arrays will enable smart
    dust WDM capabilities
  • Beam quality control (divergence) for easier
    scanning
  • Electro-optic instead of MEMs
  • Higher bit rate (will be required for very large
    networks)
  • Lower energy (about 20pJ/bit _at_ 10Mb/sec)
  • Active smart dust interfaces, robotic
    capabilities and motion

Rocket chip
UCSD
48
References
  • JM Kahn, RH Katz KSJ Pister, Emerging
    challenges mobile networking for smart dust, J.
    of Comm. and Net. 2 pp.188-196 (2000)
  • Y Song, Optical Communication Systems for Smart
    Dust, M.Sc. Thesis, Virginia polytechnic
    institute and state university, 2002
  • The following urls
  • http//www.darpa.mil/
  • http//robotics.eecs.berkeley.edu/pister/SmartDus
    t/
  • http//www-bsac.eecs.berkeley.edu/archive/users/w
    arneke-brett/SmartDust/index.html
  • http//www.xbow.com/
  • http//www.dust-inc.com/
  • http//chem-faculty.ucsd.edu/sailor/research/high
    lights.html
  • http//www.nanotech-now.com/smartdust.htm

49
thank you for your time
About PowerShow.com