Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
INSIGHT: Internet-Sensor Integration for Habitat Monitoring Murat Demirbas Ken Yian Chow Chieh Shyan Wan University at Buffalo, SUNY WSN for monitoring A sensor node (Tmote) CC2420 Radio compliant with IEEE 802.15.4 and is Zigbee ready 8MHz Texas Instruments MSP430 microcontroller (10k RAM, 48k Flash) integrated onboard antenna with 50m range indoors / 125m range outdoors integrated humidity, temperature, and light sensors (+ internal voltage) costs “in bulk” ~$5 (now $80~$130) WSN can improve Supervisory Control and Data Acquisition (SCADA) monitoring and control of a plant in industries such as telecommunications, water and waste control, energy, and transportation 2 Requirements for WSN monitoring • Energy efficiency the sensor nodes should not need batteries for at least 6 months • Remote querying and reconfiguration query data and reconfigure parameters via the Internet • Ease of deployment no pre-configuration needed • Reliability high availability, quick recovery 3 Our contributions • Remote querying basestation serves webserver and SQL database Data can be visualized, plotted, compared via webpage Email alerts based on user-defined subscriptions XML interface for data extraction • Energy-efficiency 6 months requirement met via HPL power management, delta reporting • Ease of deployment drop and play functionality via singlehop network decision • Reliability reset-timers; soft-state system • Deployment at a greenhouse 2 months deployment at UB greenhouse exposed overheating problem 4 Outline • System architecture • Energy-efficiency • Reliability • Internet-integration • Deployment results 5 System overview • Single-hop network • Basestation serves webpage access via web-browser or running an XML query • To circumvent firewall a replica is established replica obtains new data periodically via XML query 6 Basestation 7 Outline • System architecture • Energy-efficiency • Reliability • Internet-integration • Deployment results 8 HPL power management • To enable HPL sleep mode, radio is turned off after transmission • Motes wake-up 1 sec every minute for sampling and transmission 2 orders of magnitude power-saving is possible • Since motes do not need to relay transmission from more distant motes, wake-up times are kept short, and need not be coordinated 9 Delta monitoring • If the change in sensed-values between subsequent samplings are insignificant (less than delta), mote goes back to sleep without transmission originally proposed in TinyDB highly sensitive (fast-reaction) to changes in sensed values, and yet energy-efficient in the steady case scenario • In our implementation, after 20 duty cycles cumulative average readings are reported to the basestation as part of a heartbeat message, and average is reset we set delta for humidity is 1%, for temperature 0.2C, for light 2 lux, and for voltage 0.03 volts 10 Outline • System architecture • Energy-efficiency • Reliability • Internet-integration • Deployment results 11 Reset timers • Event losses might lead to livelocks in TinyOS Transmission Pending bit not being reset after transmission is done we appended a reset-timer to fix the problem • Watchdog timer to recover frozen motes if not reset by application, its overflow interrupt forces a soft reset • Watchdog timer script resets the TinyBaseStation application, the webserver and the database if they become unresponsive 12 Ease of deployment • The system can be up by just turning on all the motes and the basestation • No state is maintained at the motes in a singlehop network no coordination is needed for routing/relaying • No state is maintained at the basestation all essential applications launch automatically on startup users can locate the webpage by navigating to a dynamic DNS address MySQL stores motes information and sensor data sensor data is timestamped as it arrives in the database 13 Outline • System architecture • Energy-efficiency • Reliability • Internet-integration • Deployment results 14 Ease of use • Web-based user-interface is easy to understand • Graphical overview provides access to the data by using graphs • Tactical overview provides real-time access to the data in a top-view image • Query wizard the wizard asks a question and the user select the options desired 15 Demo http://INSIGHT.podzone.net 16 Outline • System architecture • Energy-efficiency • Reliability • Internet-integration • Deployment results 17 Deployment 18 Effects of delta monitoring • Our analysis and experimental results show a network lifetime of > 6 months Average Hourly Transmission Frequency in 24 Hours Comparison of Delta Monitoring Energy Consumption 70 3.005 60 50 2.995 Packets Max Freq Min Freq 40 2.99 Volts Packets Transmitted 3 2.985 Delta Mon., no LEDs Delta Mon., LEDs No Delta Mon., LEDs 30 2.98 20 2.975 2.97 10 2.965 0 0 1 6 11 16 21 1 2 3 4 5 6 Days Time (Hour) 19 Temperature data • Long periods of overheating (>40C) are observed • Ceiling mote recorded 2C higher temperatures than average 20 Concluding remarks • Insight simplifies high-fidelity remote querying & monitoring internet is ubiquitous users are familiar with web-browsers • Due to singlehop architecture no preconfiguration is needed no need for time sync, routing, and coordination algorithms • If a PC is already available, price is just the cost of the motes • Lifetime is around 6 months with sampling every minute 21 Future work • Integrating actuator/control mechanisms (X10?) • Using predictive monitoring to improve energy efficiency using Internet to obtain info that can help predictive monitoring • Integration with Google-Earth • An Internet-wide system for querying sensor data from Insight deployments 22