Transcript Wireless Sensing - the Internet`s Front-Tier

Wireless Sensing:
the Internet's Front-Tier
David Culler
Deborah Estrin
Federated Computing Research Conference
June 12, 2007
The Internet
The Internet Front-Tier
Low resolution Sensor, Test4, Increasing frequency
1
Acceleration (g)
0.5
0
-0.5
-1
0
2
4
6
8
10
Time (sec)
12
14
16
18
Embedded Networked Sensing
embedded
in the physical environment
(soil, canopy, rivers, groundwater, coastal)
networked
to share information and adapt function
(data, system status, control)
sensing
measurement instruments
(sensors, transducers)
an “internet” of sensors
Why “Real” Information is so Important?
Save Resources
Improve Productivity
Enable New Knowledge
Increase
Comfort
Enhance Safety & Security
Preventing Failures
High-Confidence Transport
Improve Food & H20
Protect Health
Outline
• Introduction
• Technological Foundations
• Unprecedented Information
• Participatory Sensing
• Internet Front-Tier – really
• A Broader Sense
Broad Technology Trends
Moore’s Law: # transistors on
Bell’s Law: a new computer
cost-effective chip doubles every
18 months
class emerges every 10 years
Computers
Per Person
1:106
1:103
Mainframe
Mini
Workstation
PC
Laptop
1:1
Today: 1 million transistors per $
PDA
Cell
103:1
years
Same fabrication technology provides CMOS radios
for communication and micro-sensors
Mote!
Enabling Technology
Network
Microcontroller
Flash
Storage
Radio
Communication
IEEE 802.15.4
Sensors
Enabling Systems Research
• Grand challenge visions of microscopic computing
everywhere
– Lots of Linux/Wince ARM/x86/68k + radio prototypes
• Estrin’s PC104 testbed showed that “idle listening” in
802.11 MAC dominated ALL else
• Huge emergence of interesting papers solving hypothetical
problems
 Create a platform that would expose the community to real
problems
 Share a lot of the solutions (and development overhead)
 Unconstrained by past 40 years of OS and Networking
abstractions
Silicon
Physical
World
World
WSN Research Phenomenon…
WINS
(UCLA/ROckwell)
Intel
rene’
LWIM-III
(UCLA)
SmartDust
WeC
LEAP
zeevo BT
Intel/UCB
dot
Rene
BTNodeEyes
Intel
cf-mica
trio
Mica
Telos
XBOW
mica
XBOW
rene2
Intel
MOTE2
Intel
iMOTE
XBOW
cc-dot
Bosch
cc-mica
XBOW
mica2
XBOW
micaZ
digital sun
rain-mica
Dust Inc
blue cc-TI
04
05
06
CyberPhysical
03
NETS/
NOSS
02
CENS
STC
01
NSF
00
NEST
Expedition
99
SENSIT
LWIM
DARPA
97 98
07
TinyOS 2.0
(Re)discovering the Boundaries
Over-the-air
Programming
Network
Protocols
Link
Radio
Serial
Applications and Services
Blocks,
Logs, Files
Flash
Scheduling,
Management
Streaming
drivers
MCU, Timers,
Bus,…
ADC,
Sensor I/F
WSN mote platform
Wireless
Storage
Processing
Communication Centric
Resource-Constrained
Event-driven Execution
Sensors
A worldwide community
Wireless Sensor Networks
SmartDust
Wireless
NEST
Sensors
Storage
Processing
Self-Organized Mesh Routing - nutshell
2
2
2
2
1
1
2
0
What we mean by “Low Power”
• 2 AA => 1.5 amp hours (~4 watt hours)
• Cell => 1 amp hour
(3.5 watt hours)
Cell: 500 -1000 mW
WiFi: 300 - 500 mW
GPS: 50 – 100 mW
=> few hours active
=> several hours
=> couple days
WSN: 50 mW active, 20 uW passive
450 uW => one year
45 uW => ~10 years
* System design
* Leakage (~RAM)
* Nobody fools
mother nature
Ave Power = fact * Pact + fsleep * Psleep + fwaking * Pwaking
What WSNs really look like
Client Tools
External Tools
Excel, Matlab
Enshare, etc.
GUI
Internet
Legacy
Data analysis
Embedded Network
Gateway
Field Tools
Deploy
Query
Command
Visualize
Towards the Internet Frontier –
6LoWPAN: IPv6 over IEEE 802.15.4
IEEE 802.15.4 Frame Format
D pan
Dst EUID 64
S pan
Src EUID 64
Dst16 Src16
Fchk
dsp
FCF
DSN
Len
preamble
SFD
127 bytes
Network Header
Application Data
IETF 6LoWPAN Format
01 0 0 0 0 0 1
01 0 0 0 0 1 0
Uncompressed IPv6 address [RFC2460]
HC1
Fully compressed: 1 byte
Source address
Destination address
Traffic Class & Flow Label
Next header
: derived from link address
: derived from link address
: zero
: UDP, TCP, or ICMP
40 bytes
Unprecedented Information
Science application drivers explore complex
spatial variation and heterogeneity
P. Davis, UCLA
CENS, UCLA
Dawson, UCB
Johnny Appleseed deployment myth
• Spatial sampling challenges
– Difficult to assess spatial variability and
model patterns in complex, dynamic media
(soil, water, air)
– Over-deployment not a general solution:
minimum spacing constraints, installation
difficulty, settling time
– Geometrically-determined
locations/metrics don’t capture
environment’s complex topology obstacles,
inputs (sun, precipitation, currents)
– Calls for model and data-informed
placement, iterative and adaptive
sampling
• Temporal sampling more elastic
– Many temporal signals can be fully
sampled with existing platforms
– Calls for runtime adjustments to live
within energy constraints
"Soil microorganisms mediate below- and
aboveground processes, but it is difficult
to monitor such organisms because of the
inherent cryptic nature of the soil.
Traditional 'blind' sampling methods yield
high sample variance...." [Kliornomos99]
Hansen, Harmon, Schoellhammer, et al.
Lessons from the field...
Early themes
Thousands of small devices
Minimize individual node resource needs
Exploit large numbers
Fully autonomous systems
In-network and collaborative processing
for longevity: optimize communication
New themes
Heterogeneity
Combine in situ and server processing to optimize system
Inevitable under-sampling with static sensing: mobility
Exploit multiple modalities (e.g. imagers), multiple scales
Interactivity
Coupled human-observational systems: tasking, analysis,vis.
In-network processing, system transparency for
responsiveness, data integrity, rapid-iterative deployment
Participatory sensing systems leveraging cellphone,
gps,web.
Coupled Human-Observational Systems
Transform physical observations
from batch to interactive process
• Rapid deployments are high value.
• Interactive systems take
advantage of human observation,
actuation, and inference
• Addresses critical issues such as
Slope (Spatial Analyst)
Aspect (Spatial Analyst)
Daily Average
Temperature(Geostatistic
al Analyst)
Elevation (Calculated
from Contour Map)
Aerial Photograph
(10.16cm/pixels)
Hamilton, Kaiser, Hansen, Kohler, et al.
adaptive sampling, topology
adjustment, faulty sensor
detection
• Requires real time data access,
model based analysis, system
transparency, visualization in the
field
Data, Data, Data:
Increasing role of statistical models and methods
•
Data integrity: Robust procedures for analysis in the
presence of sensing and environmental uncertainty
• Multiple scales: Designing experiments/analyses
to match observation of multi-scale phenomena
•
Opportunistic measures and models:
Integrating available measurements
with available data sources and models
Hansen, et al
Data Integrity focused on maximizing “data return”
Confidence: Tool for detecting and diagnosing
network faults with an online-variant of K-means
clustering to identify outliers in specially crafted
feature space
Model-based detection: Hidden-Markov model
complete with stochastic descriptions of system
fault learned for detection with data from NAMOS
Signatures: Short description of multivariate
probability distribution maintained for each sensor
or cluster of sensors; likelihood ratio test used to
flag readings that appear more faulty than normal
Blind calibration: New approach makes use of
projections into function spaces (smoothness classes)
Hansen, Kohler, Ramanathan,
Golubcik, Nair, Balzano
Multi-Scale terrestrial carbon fluxes
Soil CO2 concentration
Fine scale
Effects of
roots, organic
particles,
and soil
structure
Soil respiration
Plot-to field scale
Effect of group
of plants, and
gradients in
soil texture
Large-scale
Effect of
vegetation
systems, and
topography
Canopy photosynthesis
Allen, Graham, Hamilton,et al
abiotic
If you can’t go to the field with the sensor you want,
go with the sensor you have
present
future
Physical Sensors:
Microclimate above and
below ground
biotic
Chemical Sensors: gross
concentrations
•
•
•
Chemical Sensors: trace
concentrations
Acoustic and Image data
samples
Acoustic, Image sensors with
on board analysis
Organism tagging, tracking
Sensor triggered sample
collection
DNA analysis onboard
embedded device
Commercially available autonomous devices available for physical and chemical
measures only
System designs need to compensate for lack of sensor specificity, sensitivity,
availability…particularly wrt biological response variables
Leverage proxy sensors and model based signal interpretation
Imagers as biological sensors: Heartbeat of a Nestbox
Leverage context to apply
server-side, and ultimately, onboard processing to infer
“interesting behavior” (sensor
output)
Blue lines: output of automatic
image processing algorithms
applied to
cyclops images over 5 minute
intervals; Red/Green line:
temperature.
Ahmadian, Ko, Rahimi, Soatto, Estrin
Reddy, Burke, Hansen, Parker et al
Merging models and sensing
Personalized Environmental Impact Report (PEIR)
• “Footprint” calculators inform long-term choices using coarse-grained models impact
• Personalized, real-time assessment to help individuals reduce impact and minimize exposure
by viewing their own practices and habits as seen in data and inferred from models
• Employ built-in capabilities of mobile handsets to scale without specialized hardware
• Leverage model-based analyses with location traces generated using GPS, cell tower, WiFi
System components
Campaignr
N80, N95
Trace, audio, image
SensorBase
GPS-equipped mobile handset.
Custom handset software for automatic location
time-series collection, robust upload, over-the-air
upgrade/tasking, just-in-time annotation with
voice or text.
+
=
Server side tools to analyze individual spatiotemporal patterns and calculate corresponding
impact and exposure metrics to inform and advise
users.
Web-based interfaces informing and advising
users, which provide reports, real-time feedback,
visualizations and exploratory data analysis tools
for non-professional users. (For handsets and
workstations.)
Burke, Estrin, Hansen, et al
Impact / Exposure
Model
Activity type inference
Server-side
classifier
Participatory Urban Sensing: combines users, mobility, context
Enabled by
–
–
–
–
–
–
Over 2 x 109 users worldwide of cell phones.
Automated geo-coding and pervasive connectivity
Image and acoustic as data and metadata
Bluetooth connected external sensors
Local processing for data quality and triggering
Spatial interface to data and authoring
Applications
–
–
–
–
Self-administered health diagnostics
Public health/epidemiology: Water and Air
Civic concerns (transportation, safety…)
Personal Environmental Impact Report
Challenges
–
–
–
Mechanisms for selective sharing, verified location
Inference from sensor streams (gps,image,sound)
Campaign framework, data quality, incentives
participatory
sensing data
promises to
make visible
human concerns
that were
previously
unobservable…
or unacceptable
A Broader Sense
The Internet Front-Tier
Low resolution Sensor, Test4, Increasing frequency
1
Acceleration (g)
0.5
0
-0.5
-1
0
2
4
6
8
10
Time (sec)
12
14
16
18
THE Question
If Wireless Sensor Networks represent a future of “billions of
information devices embedded in the physical world,… why
don’t they run THE standard internetworking protocol?
Web Services
XML / RPC / REST / SOAP / OSGI
HTTP / FTP / SNMP
TCP / UDP
IP
Ethernet Sonet
Enet
10M
Enet
100M
Enet
1G10G
Enet
GPRS
Serial
Plugs and People
802.11
802.11a
802.11b
802.11g
RFM,CC10k,…,802.15.4
Self-Contained
Sensor Network “Networking”
Appln
Hood
EnviroTrack
FTSP
Transport
Routing
Scheduling
SPIN
TTDD
Phy
Ascent
GPSR
SPAN
ReORg
PC
Drip
Arrive
MintRoute
GRAD
GAF
FPS
Yao
SMAC
PAMAS
Link
Trickle
Deluge
MMRP
TORA
CGSR
AODVDSR
ARA
GSR
DBF
DSDV
TBRPF
Resynch
Topology
TinyDB
Diffusion
Regions
WooMac
TMAC
Pico
WiseMAC
Bluetooth
RadioMetrix
RFM
CC1000
eyes
BMAC
802.15.4
nordic
The Answer
They should
• Substantially advances the state-of-the-art in both domains.
• Implementing IP requires tackling the general case, not just a specific
operational slice
– Interoperability with all other potential IP network links
– Potential to name and route to any IP-enabled device within security domain
– Robust operation despite external factors
• Coexistence, interference, errant devices, ...
• While meeting the critical embedded wireless requirements
–
–
–
–
High reliability and adaptability
Long lifetime on limited energy
Manageability of many devices
Within highly constrained resources
Web Services
XML / RPC / REST / SOAP / OSGI
HTTP / FTP / SNMP
Proxy / Gateway
Making sensor nets make sense
LoWPAN – 802.15.4
• 1% of 802.11 power, easier to
embed, as easy to use.
• 8-16 bit MCUs with KBs, not MBs.
• Off 99% of the time
TCP / UDP
IP
Ethernet
Sonet
802.11
802.15.4, …
IETF 6lowpan
Thinking about the Physical World as “Signals”
• What is the bandwidth
of the weather?
• What is the nyquist of
the soil?
• What is the
placement noise?
• What is the sampling
jitter error?
Application drivers:
From Condition based maintenance to Precision Living…
The maturing technology will transform the
business enterprise, environmental resource
management, human interaction
– Industrial and civil infrastructure
– Individual Health and wellness
– Planet health and wellness: water, carbon,
pollution, waste
Science is our early adopter because the
technology is transformative and research
tolerates risk
Important historical precedents
- Weather modeling--early computing
- Scientific collaboration--Internet
- Experimental physics (CERN)--WWW
- Computational science--Grid
Early embedded sensing applications
• Biological and Earth Sciences
• Environmental, Civil, Bio Engineering
• Public health, Medical research
• Agriculture, Resource management
Embeddable device developments
• Energy-conserving platforms, radios
• Miniaturized, autonomous, sensors
• Standardized software interfaces
• Self-configuration algorithms
• Adaptive, iterative sampling
• Cognitive sensors
Research Ecosystem Challenges
• “Early-and-really-to-application”
–
deployments and resulting data provide feedback to system innovation… from
theory to system architecture
• Multidisciplinary research means taking turns
• Training a generation of Eco-Geeks
–
–
Mining for geeks w/diversity: gender, nationality, …
Training for the mundane and the magnificent
• Funding and sustainability
An(other) inconvenient truth
Many critical
issues facing
science,
government, and
the public call for
high fidelity and
real time
observations of
the physical world
Networks of
smart, wireless
sensors, forming
the Front-Tier of
the Internet can
help reveal the
previously
unobservable
But an
inconvenient
truth is that the
field does not
lend itself to
familiar
abstractions
and
research
practices…
Dedication
To two special people who we would have wanted
here above all …
… to give us a dozen pointed criticisms
… and two dozen wonderful new ideas.
Richard A. Newton
Jim Gray