Transcript applications

Wireless Sensors
Reliable Networks and Changing Paradigms
Kris Pister
Founder & Chief Technologist, Dust Networks
(Prof. EECS, UC Berkeley)
Dust sells reliable, low power mesh networks to OEMs
Outline
•
•
•
•
History
Applications
Standards
Technology
Berkeley Demos – 2001
Motes dropped from UAV, detect
vehicles, log and report direction
Intel Developers Forum, live demo
800 motes, 8 level dynamic network, and velocity
Seismic testing demo: real-time
data acquisition, $200 vs. $5,000 per
node
vs.
50 temperature sensors for HVAC
deployed in 3 hours. $100 vs. $800
per node.
Sensor Networks Take Off!
Industry Analysts Take Off!
800
700
Units (Millions)
600
$8.1B market for
Wireless Sensor
Networks in 2007
500
400
300
Wi-Fi nodes
Handsets
Wireless Sensor Nodes
200
100
0
2003
2004
2005
2006
2007
Source: InStat/MDR 11/2003 (Wireless); Wireless Data Research
Group 2003; InStat/MDR 7/2004 (Handsets)
Barriers to Adoption
Reliability
Standards
Ease of Use
Power consumption
Development cycles
Node size
OnWorld, 2005
0%
20%
40%
60%
80%
100%
Dust Networks
• Founded July 2002
• Focused on reliability, power consumption
• Developed TSMP
– Time Synchronized Mesh Protocol
– >99.9% reliability
– Lowest power per delivered packet
Dust Products
Product family
Features
2004
SmartMesh XR
900MHz
Reliability & low power
2005
SmartMesh XT
900 MHz, 2.4GHz
Flexibility, scale
2006
SmartMesh XD
Custom silicon
2007
IA-510
Wireless HART
Secure, reliable, robust, low power
2011
SmartMesh IP
Zigbee Green Power
SmartMesh SPOT
Next-gen silicon
Broad standards
Indoor location
50 motes, 7 hops
3 floors, 150,000sf
>100,000 packets/day
Bldg. 90
Bldg. 90
Stats on network, motes, links
Lifetime, daily, 15min
Oil Refinery – Double Coker Unit
• Scope limited to Coker
facility and support
units spanning over
1200ft
• No repeaters were
needed to ensure
connectivity
• Electrical/Mechanical
contractor installed per
wired practices
• >5 year life on C-cell
400m
Building Energy Reduction - Federspiel Controls
HVAC System Retrofits
Demonstrated Energy Savings:
• 3.7 kWh/sf/yr
• 0.34 therms/sf/yr
• Higher savings than conventional
retrofits
Building Maintenance
•
•Temperature & energy consumption monitoring
•2 hour install vs. 4 weeks for wired network
 97% reduction in installation cost
•
•
Rapid retrofit of old
buildings
Energy conservation from
modernizing systems
Platform for additional inbuilding applications
Energy Management
•Energy is the #1 cost of supermarkets
after shelf stock
•Service: monitor, analyze and reduce
power consumption
• Entire network installed
in 3 hours (vs. 3-4
days)
• Typical energy cost
reduction: 10-25%
Standardizing TSMP
• Industrial Automation Standards
– IEC 62591 (WirelessHART 2007)
– ISA100.11A
– WIA-PA (China)
• MAC is standardized in 802.15.4E (TSCH)
• Multiple network vendors: Dust, Nivis, STG, …
Barriers to Adoption
>99.9%
Wireless HART
Reliability
Standards
Ease of Use
“It just worked”
Power consumption
5-10 years
Complete networks
Development cycles
Node size
OnWorld, 2005
0%
20%
40%
60%
80%
100%
Outline
• History
• Applications
–
–
–
–
–
–
Industrial Process Automation
Commercial Building Automation
Parking management
Smart Rail
Vibration monitoring
Smart Grid
• Standards
• Technology
Emerson Process offerings, 2007
Wireless HART Architecture (from ABB)
Sampling of Wireless HART Products
 Battery
 Vibration
 Battery
 4-20 mA loop
 Solar
 Battery
 4-20 mA loop
 Thermal
Thousands of networks, 100+ countries, six continents
buildings, breweries, refineries, mines, city streets, chemical
plants, deserts, trains, steel mills, data centers, pharmaceutical
plants, offshore oil rigs…
WirelessHARTTM Adapters
ABB
Emerson
MACTek
Siemens
Adapter
THUM
BULLET
SITRANS AW200
Wheeling-Pittsburgh Steel
Need to monitor temp, coolant, lubrication
Hot slag defeated wired solutions
5% improvement in productivity (reduced maintenance)
23
Distribution of failure times
• With no sensing, need to repair frequently
• With sensing, repair near mean
– “condition-based maintenance”
Lime Kiln at Pulp & Paper Mill
• Rotating lime kiln
• Need to monitor temperature
• 5% throughput improvement
(reduced process time)
25
Distribution of completion times
• With no sensing, need to cook longer
• With sensing, mean process time = mean required
Grane Platform, North Sea
• 22 pressure sensors
• 90% reduction in
installation cost
Wireless
Sensors
27
Shell Oil, Norway
• GE Energy’s System 1 motor condition monitoring
• 200 temperature and vibration sensors
• No line power due to hazardous location rules
Wireless mesh network
1 km
2 km
Chevron’s Richmond Refinery
1 km
Richmond Refinery Wireless Umbrella
Next
• Fence monitoring
• H2S, VOC
• Location
5 km2, 90% coverage
30
Smart Building: Federspiel Controls
HVAC optimization to conserve energy
CA Tax Board savings: 459,000 kWh/yr, $55,000/yr (1 yr
payback)
No wires, no interruption to data center operations
31
Smart Cities: Streetline Networks
Wireless sensor
node
34
Urban Planning
Increasing Revenue
Finding Parking
Finding Parking
Smart Rail
• TSCH WSN enables
remote monitoring of
freight cars
• Multiple sensors per
car, every car is a
network
• Requires a strict ‘nowires’ solution, robust
enough for moving
railcars
Bearing Failure  High Cost
Vibration Monitoring
Smart Grid
Outline
• Applications
• Standards
–
–
–
–
TSMP
Zigbee
802.15.4E
IETF
• Technology
Zigbee
• The big three
– Zigbee Pro / SE1.0
– Zigbee RF4CE
• Home entertainment control
• Guarantees that cell phones will have 15.4 radios
– Zigbee IP / SE2.0
• http, TLS, DHCP, …
• Zigbee Green Power
• All use powered routers
– LPR getting little traction
• Interoperability
– AODV
– Provisioning
Protocol Integration
Application

Presentation

 Session
“other”
HTTP, SSH, Telnet, FTP
CoAP, XML,
IETF

Transport
UDP ,TCP
WSN RDP?
IPv6
RoLL RPL

Network

Data-Link

Physical
6LoWPAN
IEEE
IEEE802.3
IEEE802.11
Today’s Internet
802.15.4, 4e
802.15.4
Tomorrow’s
Internet of
Things
Evolving information flow in WSN
Business logic
Custom APP
APP
Manager
LBR
IPv6,
Proprietary
network & data fmt.
Network stack
Mote
DB
native DB fmt.
Network stack
Oski
Application
Serial API
Sensor Application
mP
Sensor
Sensor
46
Outline
•
•
•
•
History
Applications
Standards
Technology
– TSMP
– Oski
– SPOT
TSMP Foundations
• Time Synchronization
– Reliability
– Power
– Scalability
• Reliability
– Frequency diversity
• Multi-path fading, interference
– Spatial diversity
• True mesh (multiple paths at each hop)
– Temporal diversity
• Secure link-layer ACK
• Power
– Turning radios off is easy
Power-optimal communication
• Assume all motes share a network-wide synchronized
sense of time, accurate to ~1ms
• For an optimally efficient network, mote A will only
be awake when mote B needs to talk
A
A wakes up and listens
B
B transmits
B receives ACK
A transmits ACK
Expected packet start time
Worst case A/B clock skew
Packet transmission and acknowledgement
Mote
Current
Radio TX startup
Packet TX
Radio TX/RX turnaround
(2011): 15 mC
ACK RX
(2008): 50 mC
Charge cost (2003): 300 mC
Idle listen (no packet exchanged)
Mote
Current
Radio RX startup
Empty receive
(2011): 5 mC
(2008): 27 mC
Charge cost (2003): 70 mC
Mesh Networking
IP
Gateway
IEEE 802.15.4 Mote
Sensor
• 802.15.4 PHY, 2.4 GHz
• Time Synchronized for low power & scalability
– All nodes run on batteries, for 5-10 years
• Channel Hopping and full mesh for reliability
– 99.999% “best effort” packet delivery
Time Synchronization
• Timestamps available to application
– RMS <0.1ms with Wireless HART
– Substantially lower with SmartMesh IP
Absolute time synch
Stratum 1 server
NTP
PM or LM
Mote
• Relative error: 0.1ms avg., 1ms max
• Absolute error on PM:
• 0.3ms avg. ; 99.9% <1ms; 10ms worst case
• 1us w/ 1588
54
Oski
• Future-proof horsepower
– 32 bit ARM Cortex M3
– 512kB flash, 72kB RAM
• Revolutionary radio & network
–
–
–
–
IPv6 router < 20μA
10 years with an AA lithium battery
Microsecond timestamps
Location
• Fast application development
• Multi-protocol routing
– 6LoWPAN
– Zigbee SE 1, 2; Pro
– HART
Mote-on-chip current vs. sample date
RX Current
0dBm TX Current
45
45
40
40
35
35
CEL
30
25
Ember
MSP430
+CC2420
I_TX [mA]
I_RX [mA]
30
TI
20
Freescale
TI
10
10
Dust Networks
Dust Networks
5
5
Jan-06
Jan-07
Jan-08
Sampling date
Jennic
Jennic
15
Jan-05
Ember
20
15
0
Jan-04
Freescale
25
Jan-09
Jan-10
Jan-11
Jan-12
0
Jan-04
Jan-05
Jan-06
Jan-07
Jan-08
Jan-09
Jan-10
Jan-11
Sampling date
56
Jan-12
Real-Time Location Systems
• RTLS costs often dominated by infrastructure
– Power and/or data cabling for readers
• Barrier to initial deployment
Electrical Installation (40%)
Sensory Hardware (25%)
Project Management (9%)
Tools, Equipment and Consumables
(7%)
Solar equipment (4%)
Surveying (0.5%)
Contingency (10% of total)
Engineering & Drafting (11%)
Lifts, Operators (4%)
57
SmartMesh SPOT
Asset Management
System & Location Engine
Network
Manager
Locn: Room
327, west wall
Fixed Battery
Powered Mote
27.2m
22.5m
40.1m
Mobile
Mote
17.8m
23.2m
Sensor
58
58
Indoor localization accuracy
102-point West Dust Offices 110218
1
0.9
0.8
Linear (a = 0.59)
Nonlin W SE (a = 0.60)
0.7
CDF
0.6
0.5
0.4
0.3
0.2
0.1
0
0
1
2
3
4
5
Error [m]
6
7
8
9
10
SmartMesh SPOT Advantages
• No site survey
– Field-proven, self-forming, self-healing TSCH
mesh
• No wires
– Battery/scavenger-powered “peel-and-stick”
infrastructure
• …and a true IP network
– Sensors: button, temp, shock, …
– Outputs: displays, alarms, …
Summary
• Real applications exist
• Standards are a reality
• Existing products have high reliability and low
power
• Low-infrastructure localization is coming
• Result:
– Work on applications of WSN, not sensor
networking