Wireless Sensor Networks for Habitat Monitoring Alan Mainwaring
Wireless Sensor Networks for Habitat Monitoring Alan Mainwaring, Joseph Polastre, Robert Szewczyk, and David Culler Intel Research Lab. / UCBerkely Seo, Dong Mahn
Contents Introduction l Application Requirements l System Architecture l Design and Implementation Strategies l Sensor Network Services l Current Progress l Additional materials l Conclusion l September 8, 2005 2
Introduction l l Habitat and environmental monitoring Technical interests in these applications developing an appropriate sensor network architecture l simple, concrete solutions l application-driven approach l l actual problems from potential ones l relevant issues from irrelevant ones l collaboration with scientists in other fields September 8, 2005 3
Introduction (cont. ) l Instrumentation of natural spaces with networked sensors long-term data collection at scales l localized measurements l detailed information l integration of on-board processing, local storage, networking l l complex filtering and triggering functions l application- and sensor-specific data compression algorithms September 8, 2005 4
Introduction (cont. ) l complete integration l l produces smaller, low-power devices increased power efficiency flexibility low-power radios with well-designed protocol A specific habitat monitoring application l l l collection of requirements, constraints and guidelines basis for the resulting sensor network architecture in the realworld hardware and sensor platforms patch gateways, basestations and databases design and implementation of the essential network services l power management, communications, retasking and node management September 8, 2005 5
Application Requirements A. Field Stations and Research Overviews l Great Duck Island (GDI) l l l 44. 09 N, 68. 15 W, 237 acre, State of Maine focus on basic ecology, large breeding colonies of Leech’s Strom Petrels and other seabirds basic environmental parameters l light, temperature, humidity, pressure entrance/exit events the James San Jacinoto Mountains Reserve (JMR) l l l 33. 48 N, 116. 46 W, 29 acre, California NSF Center : sensing infrastructures, multimedia sensors monitoring ecosystems l l response of vegetation to climate changes acoustical sensing of birds for identification, estimation populations September 8, 2005 6
Application Requirements (cont. ) B. General application requirements 1) Internet access 2) Hierarchical network l l Field stations need host Internet connectivity and database systems Habitats are located up to several kilometers multiple patches of sensor networks 3 to 4 patches of 100 static (not mobile) nodes 3) Sensor network longevity l l run for 9 months from non-rechargeable power sources multiple field seasons September 8, 2005 7
Application Requirements (cont. ) 4) Operating off-the-grid l l operate with bounded energy supplies renewable energy 5) Management at-a-distance l l to monitor and manage sensor networks over the Internet except for installation and removal of nodes 6) Inconspicuous operation l should not disrupt the natural processes or behaviors 7) System behavior l SNs exhibit stable, predictable, and repeatable behavior September 8, 2005 8
Application Requirements (cont. ) 8) In-situ interactions l Local interactions l l initial deployment, maintenance tasks PDA l query a sensor, adjust operational parameters, or simply assist in location devices 9) Sensors and sampling l l light, temperature, infrared, relative humidity, barometric pressure acceleration/vibration, weight, chemical vapors, gas concentrations, p. H, noise levels September 8, 2005 9
Application Requirements (cont. ) C. Data models l l l Archiving sensor readings for offline data mining and analysis logs to databases in the wired, powered infrastructure nodal data summaries, periodic health-and-status monitoring September 8, 2005 10
System Architecture l lowest lever of the sensing application l autonomous sensor nodes l general purpose computational module l l l programmable unit computation, storage, bidirectional communication with analog and digital sensors 2 advantages from traditional data logging systems l can be retasked, can easily communicate application-specific sensing module smaller and cheaper individual sensors l l higher robustness cooperation multihop network, forwarding each other’s messages in-network aggregation September 8, 2005 11
System Architecture (cont. ) l Sensor Gateway l l l each sensor patch communicate with the sensor network and provides commercial WLAN AP is co-located with the base station additional computation and storage enough energy from a car battery Base Station l l power, housing communicates with the sensor patch via WLAN WAN, persistent data storage “custody transfer” model : SMTP messages, bundles September 8, 2005 12
System Architecture (cont. ) l User interaction l access the replica of the base station database l easy l integration with data analysis and mining tools remote control of the network l PDA-sized September 8, 2005 device, gizmo 13
System Architecture (cont. ) September 8, 2005 14
System Architecture (cont. ) Patch Network Sensor Node Sensor Patch Gateway Transit Network Client Data Browsing and Processing Basestation Base-Remote Link Internet Data Service September 8, 2005 15
Design and Implementation Strategies A. Sensor Network Node l l l UC Berkely motes, MICA single channel, 916 MHz radio, 40 kbps Atmel Atmega 103 microcontroller running at 4 MHz 512 KB nonvolatile storage 2 AA batteries, DC boost converter September 8, 2005 16
Design and Implementation Strategies (cont. ) B. Sensor Board l l l environmental monitoring sensor board Mica Weather Board barometric pressure module l l humidity sensor l l 0. 1 mbar from 300 to 1100 mbar 1 picofarad (± 3% relative humidity) thermopile, passive infrared sensor photoresistor, temperature unique combination of sensors l variety of aggregate operations September 8, 2005 17
Design and Implementation Strategies (cont. ) l I 2 C analog to digital converter l l 8 by 8 power switch interoperability l 51 pin expansion connector September 8, 2005 18
Design and Implementation Strategies (cont. ) C. Energy budget l l l run for 9 months, 2 AA batteries 2200 m. Ah at volts, 8, 148 m. Ah per day sleep state l l turning off sensors, radio, putting processor into sleep mode modify Mica motes with a Schottky diode September 8, 2005 19
Design and Implementation Strategies (cont. ) D. Electro-mechanical Packaging l to protect the device, weather-proofing E. Patch Gateways l l Cerf. Cube, Strong. Arm-based embedded system Compact. Flash-based 802. 11 b Linux, IBM Micro. Drive up to 1 GB Solar panel F. Base-station installation l l JMR : T 1 line, GDI : two-way satellite connetion turnkey system September 8, 2005 20
Design and Implementation Strategies (cont. ) G. Database Management System l l l Postgres SQL database time-stamped reading from sensors health status of individual sensors network metadata H. User Interfaces l l l GIS systems, statistics and data analysis packages powerful interfaces to relational databases web based interface, gizmo September 8, 2005 21
Design and Implementation Strategies (cont. ) Satellite router WWW power strip Web power strip 4 -port VPN router and 16 -port Ethernet switch DB Power over LAN midspan IR Burrow Camera #1 IR Burrow Camera #5 IR Burrow Camera #2 Sensor Patch. IR Burrow Camera #6 IR Burrow Camera #3 ) IR Burrow Camera #4 IR Burrow Camera #7 IBM laptop #1 IR Burrow Camera #8 Power over LAN Midspan Wireless bridge IBM laptop #2 DB Northern WAP Ethernet switch 110 VAC service Burrow Camera Configuration 12 VDC, 0. 9 A Southern WAP Mica 2 -EPRB#2 Wireless bridge September 8, 2005 network Axis 2401 Video Server Mica 2 -EPRB#2 Axis 2130 PTZ South 12 V Po. L Active Splitter 916 MHz 22
Sensor Network Services A. Data sampling and collection l l cost of data processing and compression against cost of data transmission each packet 25 bytes September 8, 2005 23
Sensor Network Services (cont. ) B. Communications l l hardware and a set of routing and media access algorithms GAF (Geographic Adaptive Fidelity), SPAN September 8, 2005 24
Sensor Network Services (cont. ) l proposed approaches for scheduled communication routing tree set each mote’s lever form gateway schedule nodes sleep state following level is awaken and packets are relayed until completed entire network return to sleep mode l path or subtree l initial l low power MAC protocol l S-MAC, Aloha l turning off radio during idle periods September 8, 2005 25
Sensor Network Services (cont. ) C. Network Retasking l to adjust the functionality of individual nodes l l duty cycle, sampling rates … tiny virtual machine, Maté D. Health and Status Monitoring l l l monitoring the mote’s health and the health of neighboring motes Health and monitoring messages sent to the gateway not reliable transport, low latency, infrequently September 8, 2005 26
Current Progress l deployed l l l plan to add an intermediate tier of WLAN need calibration or auto-calibration procedure current focus l l l two small scale sensor networks in JMR and GDI all core architecture components energy efficient strategies for multihop routing will evaluate intention l l l to develop and package a habitat monitoring kit will be completed in 6 months goal is to tackle the technical problems and to meet the application requirements set September 8, 2005 27
Additional Materials l Node architecture advances l Problems observed in previous deployment l l l Size – motes were too large to fit in many burrows Packaging – did not provide adequate protection for electronics or proper conditions for sensors Reliability – last retreat talk; high rate of node loss, lack of scientifically meaningful environmental data Power consumption – boost converter a minimal return at a high price New generation of motes to address most of these concerns l l l Platform based on mica 2 dot Primarily calibrated, digital sensors Multiple application-specific packaging, power, and sensing options September 8, 2005 28
Additional Materials (cont) September 8, 2005 29
Additional Materials (cont) September 8, 2005 30
Additional Materials (cont) l Miniature weather station l Sensor suite l l l Power supply l l Sensirion humidity + temperature sensor Intersema pressure + temperature sensor TAOS total solar radiation sensor Hamamatsu PAR sensor Radiation sensors measure both direct and diffuse radiation SAFT Li. S 02 battery, ~1 Ah @ 2. 8 V Packaging l l HDPE tube with coated sensor boards on both ends of the tube Additional PVC skirt to provide extra shade and protection against the rain September 8, 2005 31
Additional Materials (cont) l Burrow occupancy detector l Sensor suite l l l Power supply l l l Sensirion humidity + temperature sensor Melexis passive IR sensor + conditioning circuitry Great. Batch lithium thionyl chloride 1 Ah battery Maxim 5 V boost converter for Melexis circuitry Packaging l Sealed HDPE tube, emphasis on small size September 8, 2005 32
Additional Materials (cont) l Software architecture advances l Bi-directional communication with low-power listenting l l Improved power management scheme l l 0. 1% duty cycle Parameter adjustment and query Sample rate changes, sensor status queries Fine granularity through Std. Control interface 20 u. A sleep mode Multihop deployment planned for July What it isn’t: GSK l l Emphasis on simplicity and reliability, rather than generality Compatible with most GSK server-side interfaces September 8, 2005 33
Additional Materials (cont) l Application status l Sensor network l l 26 burrow motes deployed 12 weather station motes deployed (+2 for monitoring the insides of the base station case) l l Redundant database setup online l l l Another 14 are awaiting deployment within days 2 base stations logging packets to 2 database servers Replication to Berkeley Verification infrastructure l l l Overview cameras in place Burrow cameras temporarily offline, wireless bridge problem Video logging still needs to be synchronized with the mote data service September 8, 2005 34
Additional Materials (cont) l Packaging evaluation l We observed what happens to motes when packaging fails l l Battery venting, H 2 SO 3 corroding the entire mote Need to assemble the package correctly – we failed to create proper indication os a good seal Majority of packages survived severe weather! Still awaiting evaluation whether the package creates an environment suitable for sensing l Convective heating, etc. September 8, 2005 35
Additional Materials (cont) September 8, 2005 36
Additional Materials (cont) September 8, 2005 37
Additional Materials (cont) September 8, 2005 38
Additional Materials (cont) September 8, 2005 39
Additional Materials (cont) September 8, 2005 40
Additional Materials (cont) September 8, 2005 41
Additional Materials (cont) l http: //www. jamesreserve. edu/ September 8, 2005 42
Additional Materials (cont) l http: //www. greatduckisland. net/ September 8, 2005 43
Conclusion l Habitat and environmental monitoring l l collaborating with l l l important class of sensor network applications College of the Atlantic and the James Reserve low-level energy constraints of the sensor nodes data delivery requirements energy budget Tight energy bounds and the need for predictable operation guide the development of application architecture and services. September 8, 2005 44
Reference l l l l http: //www. jamesreserve. edu/ http: //www. greatduckisland. net/ Robert Szewczyk, Joe Polastre, Alan Mainwaring, “Fresh from the boat: Great Duck Island habitat monitoring”, June 18, 2003 Alan Mainwaring, Joseph Polastre, Robert Szewczyk, David Culler, John Anderson, “Wireless Sensor Networks for Habitat Monitoring”, ACM WSNA’ 02, September 28, 2002, Atlanta, Georgia, USA. Joseph Robert Polastre, “Design and Implementation of. Wireless Sensor Networks for Habitat Monitoring” Kemal Akkaya, Mohamed Younis, “A Survey on Routing Protocols for Wireless Sensor Networks” Wei Hong, “Overview of the Generic Sensor Kit (GSK)” Robert Szewczyk, “Application-driven research on Tiny. OS platform” September 8, 2005 45
September 8, 2005 46
- Slides: 46