Life Under your Feet: A Wireless Soil Ecology Sensor Network

Transcription

1 Life Under your Feet: A Wireless Soil Ecology Sensor Network R. Musaloiu-E., A. Terzis, K. Szlavecz, A. Szalay *, J. Cogan *, J. Gray Computer Science Department, JHU Earth and Planetary Sciences Department, JHU * Physics Department, JHU Microsoft Research EmNets 2006 Life Under your Feet 1

2 In the Beginning The Cast An ecologist: Katalin Szlavecz An undergrad student working in a summer project: Joshua Cogan The Stage The urban forest near the JHU campus Off-the-self soil moisture and temperature sensors A few Mica2 motes The Goal Test whether WSNs can be used to collect soil monitoring data at spatial and temporal scales larger than that of data loggers EmNets 2006 Life Under your Feet 2

3 First Step The sensors were thoroughly tested in the lab 2 out of 6 sensors misbehaved Relatively precise (+/- 0.5 C) but offset of 1.5 C from a NIST approved thermocouple. Next step: teach the motes to play their part. EmNets 2006 Life Under your Feet 3

4 Second Step More actors: CS faculty (Andreas Terzis) CS grad student (Razvan Musaloiu-E.) DB experts: Alex Szalay and Jim Gray More setup: MicaZ motes MSSQL Skyserver DB software The goal Build an end-to-end data collection system EmNets 2006 Life Under your Feet 4

5 System requirements Measurement fidelity All the raw measurements should be collected and persistently stored Measurement accuracy and precision Temperature data error < 0.5 C, volumetric moisture data error < 1%. Fusion with external sources Comparing measurements with external data sources is crucial Additional Requirements Sampling Frequency: minutes Experiment Duration: years Deployment Size: networks should be deployed in ways that capture the heterogeneity of land use EmNets 2006 Life Under your Feet 5

6 System Architecture EmNets 2006 Life Under your Feet 6

7 Sensor Node Hardware Sensors: Watermark soil moisture Irrometer soil thermistor Onboard Temperature and Light MicaZ motes Waterproof cases, cables, connectors,... EmNets 2006 Life Under your Feet 7

8 EmNets 2006 Life Under your Feet 8

9 EmNets 2006 Life Under your Feet 9

10 Mote Software Mote Base Mote Base status message data request data message 2min1s... Bulk Phase Send and Wait Phase... Receiving Sending (a) (b) Sensors are sampled every minute Raw measurements are stored in flash Storage overrun in > 20 days Data retrieved using NACK-based sliding window protocol Each mote sends every 2 min a status messages to basestation: (#meas, bat voltage) Web-accessible real-time monitoring EmNets 2006 Life Under your Feet 10

11 Database Design DB Design derived from experiment and WSN setup Self-documenting framework Raw data follow a multistage validation process Stored data available through Web pages and Web Services Datacube for analysis 7M datapoints collected EmNets 2006 Life Under your Feet 11

12 Calibration workflow Calibrations in the Lab Mote Resistor Calibration Temperature Sensor Calibration Temperature sensor A/D units Voltage Resistance Soil Temperature Reference voltage A/D units Voltage Moisture Sensor Calibration Moisture sensor A/D units Voltage Resistance Water Deficit Soil Matrix Potential Light Intensity A/D units Air Temperature A/D units Temperature Conversion Soil Water Potential-> Volumetric Conversion CPU clock UTC DateTime Air Temperature Celsius Water Content Volumetric EmNets 2006 Life Under your Feet 12

13 Deployment Site 2m 2m 8m 10 motes deployed on a slanted grid, ~2m apart 147 days of deployment Basestation within direct radio range Some links had unexpectedly poor quality Had to collect some of the data using laptop EmNets 2006 Life Under your Feet 13

14 Ecology Results (Temperature) EmNets 2006 Life Under your Feet 14

15 Ecology Results (Soil Moisture) EmNets 2006 Life Under your Feet 15

16 Energy consumption Motes go to sleep when not collecting data/sending reports Easy to calculate average current draw Powered by AA alkaline batteries Linear discharge curve Possible to predict node lifetime EmNets 2006 Life Under your Feet 16

17 Data Transmission Performance Good Link Lossy Link EmNets 2006 Life Under your Feet 17

18 Problems we encountered Waterproofing the boxes Moisture sensor behavior in the wild Power consumption Data provenance Reprogramming EmNets 2006 Life Under your Feet 18

19 Conclusions Sensor Networks are still expensive in time, money and know-how Need for data analysis tools Need for increased confidence in collected measurements End-to-End system that transforms raw measurements to scientifically significant data EmNets 2006 Life Under your Feet 19

20 Questions? For more information EmNets 2006 Life Under your Feet 20