Revealing the hidden lives of underground animals with - - PowerPoint PPT Presentation
Revealing the hidden lives of underground animals with Magneto-Inductive tracking Andrew Markham Stephen Ellwood Niki Trigoni David Macdonald Computing Laboratory Wildlife Conservation Research Unit University of Oxford University of Oxford
Andrew Markham Niki Trigoni Stephen Ellwood David Macdonald
Computing Laboratory University of Oxford
Wildlife Conservation Research Unit University of Oxford 5 November 2010: ACM SenSys
Motivation Steps in MI localization System Design Results Limitations and Lessons learnt Summary
1 Motivation 2 Steps in MI localization 3 System Design 4 Results 5 Limitations and Lessons learnt 6 Summary
Andrew Markham SenSys 2010 2/42
Motivation Steps in MI localization System Design Results Limitations and Lessons learnt Summary
Tracking animals above ground is relatively well researched
ZebraNet - GPS-WSN Virtual Fencing - GPS-WSN TurtleNet - GPS-WSN WildSensing - RFID-WSN ...
Andrew Markham SenSys 2010 3/42
Motivation Steps in MI localization System Design Results Limitations and Lessons learnt Summary
Example burrowing species: badgers Nocturnal medium sized carnivores (8 kg) Live in extensive (20 m x 10 m) setts Tunnels between 1 and 3 m deep typically 1 to 20 badgers in a sett
Andrew Markham SenSys 2010 4/42
Motivation Steps in MI localization System Design Results Limitations and Lessons learnt Summary
How can we track animals underground? Radio + Soil = no signal Digging out a den destroys it Ground penetrating radar (GPR) does not indicate positions of animals
Andrew Markham SenSys 2010 5/42
Motivation Steps in MI localization System Design Results Limitations and Lessons learnt Summary
Magnetic fields are unaffected by:
Soil Water Air Vegetation
Magnetic fields are affected by metal, but luckily, there is not much of that in ancient woodlands!
Andrew Markham SenSys 2010 6/42
Motivation Steps in MI localization System Design Results Limitations and Lessons learnt Summary
1 Generation of magnetic fields 2 Sensing and compression 3 Information transfer 4 Localization
Andrew Markham SenSys 2010 7/42
Motivation Steps in MI localization System Design Results Limitations and Lessons learnt Summary
1 Generation of magnetic fields 2 Sensing and compression 3 Information transfer 4 Localization
Compare with well known RF RSSI based techniques
Andrew Markham SenSys 2010 7/42
Motivation Steps in MI localization System Design Results Limitations and Lessons learnt Summary
Andrew Markham SenSys 2010 8/42
Motivation Steps in MI localization System Design Results Limitations and Lessons learnt Summary
Andrew Markham SenSys 2010 9/42
Motivation Steps in MI localization System Design Results Limitations and Lessons learnt Summary
Spatial distribution of field is more complex for MI RSSI
RSSI(x, y, z) = α0 + α1 log(r)
MI
Bz (x, y, z) = µ0I 4π
4
[ (−1)ada ra[ra + (−1)a+1ca] − Ca ra[ra + da] ] Bx (x, y, z) = µ0I 4π
4
[ (−1)a+1z ra[ra + da] ] By (x, y, z) = µ0I 4π
4
[ (−1)a+1z ra[ra + (−1)a+1ca] ] |B| =
Andrew Markham SenSys 2010 10/42
Motivation Steps in MI localization System Design Results Limitations and Lessons learnt Summary
Spatial distribution of field is more complex for MI Can’t be treated as point source RSSI
RSSI(x, y, z) = α0 + α1 log(r)
MI
Bz (x, y, z) = µ0I 4π
4
[ (−1)ada ra[ra + (−1)a+1ca] − Ca ra[ra + da] ] Bx (x, y, z) = µ0I 4π
4
[ (−1)a+1z ra[ra + da] ] By (x, y, z) = µ0I 4π
4
[ (−1)a+1z ra[ra + (−1)a+1ca] ] |B| =
Andrew Markham SenSys 2010 10/42
Motivation Steps in MI localization System Design Results Limitations and Lessons learnt Summary
Magnetic fields fall off more rapidly RSSI
RSSI ∝ 1 r 2
40 dB/decade MI
|B| ∝ 1 r 3
60 dB/decade
Andrew Markham SenSys 2010 11/42
Motivation Steps in MI localization System Design Results Limitations and Lessons learnt Summary
Magnetic fields fall off more rapidly Range less than traditional radio RSSI
RSSI ∝ 1 r 2
40 dB/decade MI
|B| ∝ 1 r 3
60 dB/decade
Andrew Markham SenSys 2010 11/42
Motivation Steps in MI localization System Design Results Limitations and Lessons learnt Summary
Magnetic field controlled by size and shape of generating coil RSSI
10 5 5 10 10 5 5 10
MI
10 5 5 10 10 5 5 10
Andrew Markham SenSys 2010 12/42
Motivation Steps in MI localization System Design Results Limitations and Lessons learnt Summary
Magnetic field controlled by size and shape of generating coil RSSI
10 5 5 10 10 5 5 10
MI
10 5 5 10 10 5 5 10
Andrew Markham SenSys 2010 12/42
Motivation Steps in MI localization System Design Results Limitations and Lessons learnt Summary
Magnetic field controlled by size and shape of generating coil RSSI
10 5 5 10 10 5 5 10
MI
10 5 5 10 10 5 5 10
Andrew Markham SenSys 2010 12/42
Motivation Steps in MI localization System Design Results Limitations and Lessons learnt Summary
Magnetic field controlled by size and shape of generating coil RSSI
10 5 5 10 10 5 5 10
MI
10 5 5 10 10 5 5 10
Andrew Markham SenSys 2010 12/42
Motivation Steps in MI localization System Design Results Limitations and Lessons learnt Summary
Magnetic field controlled by size and shape of generating coil Simple to alter field patterns to optimize localization RSSI
10 5 5 10 10 5 5 10
MI
10 5 5 10 10 5 5 10
Andrew Markham SenSys 2010 12/42
Motivation Steps in MI localization System Design Results Limitations and Lessons learnt Summary
MI penetrates any non-metallic objects and does not suffer from multipath RSSI
0.02 0.04 0.06 0.08 0.1
Distance
6 7 8 9 10 11 12 13 14
RSSI
MI
0.02 0.04 0.06 0.08 0.1
Distance
6 7 8 9 10 11 12 13 14
B-field
Andrew Markham SenSys 2010 13/42
Motivation Steps in MI localization System Design Results Limitations and Lessons learnt Summary
MI penetrates any non-metallic objects and does not suffer from multipath Environmental obstacles do not affect MI localization accuracy RSSI
0.02 0.04 0.06 0.08 0.1
Distance
6 7 8 9 10 11 12 13 14
RSSI
MI
0.02 0.04 0.06 0.08 0.1
Distance
6 7 8 9 10 11 12 13 14
B-field
Andrew Markham SenSys 2010 13/42
Motivation Steps in MI localization System Design Results Limitations and Lessons learnt Summary
Andrew Markham SenSys 2010 14/42
Motivation Steps in MI localization System Design Results Limitations and Lessons learnt Summary
MI sensor detects signals from three orthogonal transponders Simultaneously measures RSSI Vector magnitude taken to ensure rotational invariance |B| =
Andrew Markham SenSys 2010 15/42
Motivation Steps in MI localization System Design Results Limitations and Lessons learnt Summary
MI sensor operates at low frequency (125 kHz) and is much more energy efficient than RF RSSI
MI
Andrew Markham SenSys 2010 16/42
Motivation Steps in MI localization System Design Results Limitations and Lessons learnt Summary
MI sensor operates at low frequency (125 kHz) and is much more energy efficient than RF 100 times less energy - allows for continuous tracking RSSI
MI
Andrew Markham SenSys 2010 16/42
Motivation Steps in MI localization System Design Results Limitations and Lessons learnt Summary
Underground Aboveground 0.05 0.1 0.15 0.2 0.25 0.3
Average current (mA)
Processor Data Beacons Flash MI Sensor
Andrew Markham SenSys 2010 17/42
Motivation Steps in MI localization System Design Results Limitations and Lessons learnt Summary
Note that the majority of energy is used by the CPU, not by localization
Underground Aboveground 0.05 0.1 0.15 0.2 0.25 0.3
Average current (mA)
Processor Data Beacons Flash MI Sensor
Andrew Markham SenSys 2010 17/42
Motivation Steps in MI localization System Design Results Limitations and Lessons learnt Summary
Each collar typically receives 250 000 readings per day
Andrew Markham SenSys 2010 18/42
Motivation Steps in MI localization System Design Results Limitations and Lessons learnt Summary
Each collar typically receives 250 000 readings per day 7 bytes per record (antenna id, timestamp, strength) = 1.7 Mbytes per day
Andrew Markham SenSys 2010 18/42
Motivation Steps in MI localization System Design Results Limitations and Lessons learnt Summary
Each collar typically receives 250 000 readings per day 7 bytes per record (antenna id, timestamp, strength) = 1.7 Mbytes per day Break into streams for each antenna = 300 kbytes per day
Andrew Markham SenSys 2010 18/42
Motivation Steps in MI localization System Design Results Limitations and Lessons learnt Summary
Each collar typically receives 250 000 readings per day 7 bytes per record (antenna id, timestamp, strength) = 1.7 Mbytes per day Break into streams for each antenna = 300 kbytes per day Can reduce further - lots of similarity between subsequent readings
Andrew Markham SenSys 2010 18/42
Motivation Steps in MI localization System Design Results Limitations and Lessons learnt Summary
Haar Wavelet Transform Elias-Gamma Encoding
0: 1 1: 010 2: 011 3: 00100 4: 00101 5: 00110 6: 00111 7: 0001000
Reconstruction at basestation
Andrew Markham SenSys 2010 18/42
Motivation Steps in MI localization System Design Results Limitations and Lessons learnt Summary
Haar Wavelet Transform Elias-Gamma Encoding
0: 1 1: 010 2: 011 3: 00100 4: 00101 5: 00110 6: 00111 7: 0001000
Reconstruction at basestation
Andrew Markham SenSys 2010 18/42
Motivation Steps in MI localization System Design Results Limitations and Lessons learnt Summary
Each collar typically receives 250 000 readings per day 7 bytes per record (antenna id, timestamp, strength) = 1.7 Mbytes per day Break into streams for each antenna = 300 kbytes per day Can reduce further - lots of similarity between subsequent readings 50 to 100 kbytes per day
Andrew Markham SenSys 2010 18/42
Motivation Steps in MI localization System Design Results Limitations and Lessons learnt Summary
Andrew Markham SenSys 2010 19/42
Motivation Steps in MI localization System Design Results Limitations and Lessons learnt Summary
Latency of many conventional WSN protocols too high Need to bulk transfer large amounts of data (100 kbyte) in a very short contact window (minutes) Simple high speed reliable data transfer with multiple channels Periodic beacons for localization above-ground Overall duty cycle 0.3%
Andrew Markham SenSys 2010 20/42
Motivation Steps in MI localization System Design Results Limitations and Lessons learnt Summary
10 5 5 10 [m] 10 5 5 10 [m]
Andrew Markham SenSys 2010 21/42
Motivation Steps in MI localization System Design Results Limitations and Lessons learnt Summary
Field patterns for MI are highly non-linear RSSI
2 4 6 8 10 Distance [m] 10 20 30 40 50 60 Field strength [dB]
MI
2 4 6 8 10 Distance [m] 10 20 30 40 50 60 Field strength [dB]
Andrew Markham SenSys 2010 22/42
Motivation Steps in MI localization System Design Results Limitations and Lessons learnt Summary
Field patterns for MI are highly non-linear Use non-linear least squares methods RSSI
2 4 6 8 10 Distance [m] 10 20 30 40 50 60 Field strength [dB]
MI
2 4 6 8 10 Distance [m] 10 20 30 40 50 60 Field strength [dB]
Andrew Markham SenSys 2010 22/42
Motivation Steps in MI localization System Design Results Limitations and Lessons learnt Summary
1 Motivation 2 Steps in MI localization 3 System Design 4 Results 5 Limitations and Lessons learnt 6 Summary
Andrew Markham SenSys 2010 23/42
Motivation Steps in MI localization System Design Results Limitations and Lessons learnt Summary
Zoologists interested in tracking animals underground posed the following requirements: Determine 3-D location within 50 cm Operate to a depth of at least 4 m Temporal resolution ∼ 1 s 3 month lifetime Lightweight
Andrew Markham SenSys 2010 24/42
Motivation Steps in MI localization System Design Results Limitations and Lessons learnt Summary
Generates time-multiplexed digitally modulated signals 125 kHz carrier frequency, Manchester coded 1300 bps Antennas act as resonant LC tanks
Antenna ID (2 bytes) Timestamp (4 bytes) CRC (1 byte) Andrew Markham SenSys 2010 25/42
Motivation Steps in MI localization System Design Results Limitations and Lessons learnt Summary
Custom made tracking tags Atmel AVR-Zigbit Microcontroller and Transceiver AS3932 Analog Front End (AFE) 4 Mbyte Serial Flash Lithium thionyl chloride battery
Andrew Markham SenSys 2010 26/42
Motivation Steps in MI localization System Design Results Limitations and Lessons learnt Summary
Basestation with SD storage T-mote Skys as above-ground ‘detection’ nodes
Andrew Markham SenSys 2010 27/42
Motivation Steps in MI localization System Design Results Limitations and Lessons learnt Summary
1 Motivation 2 Steps in MI localization 3 System Design 4 Results 5 Limitations and Lessons learnt 6 Summary
Andrew Markham SenSys 2010 28/42
Motivation Steps in MI localization System Design Results Limitations and Lessons learnt Summary
5 10
Distance [m]
10 20 30 40 50 60
Magnitude of magnetic field [dB] Measured Predicted
Andrew Markham SenSys 2010 29/42
Motivation Steps in MI localization System Design Results Limitations and Lessons learnt Summary
1 1 2 3 4 x [m] 1.0 0.5 0.0 0.5 1.0 1.5 2.0 y [m]
Ground truth Coil positions Estimated locations
Andrew Markham SenSys 2010 30/42
Motivation Steps in MI localization System Design Results Limitations and Lessons learnt Summary
4 adult badgers trapped in January 4 large loop antennas (4m × 1 m) installed around sett
Andrew Markham SenSys 2010 31/42
Motivation Steps in MI localization System Design Results Limitations and Lessons learnt Summary
−80 −60 −40 −20 20 40 60
x [m]
−30 −20 −10 10 20 30 40 50
y [m]
Surface nodes Holes
Andrew Markham SenSys 2010 32/42
Motivation Steps in MI localization System Design Results Limitations and Lessons learnt Summary
−80 −60 −40 −20 20 40 60
x [m]
−30 −20 −10 10 20 30 40 50
y [m]
Surface nodes Holes
Andrew Markham SenSys 2010 32/42
Motivation Steps in MI localization System Design Results Limitations and Lessons learnt Summary
−2 2 4 6 8 10 12
x [m]
−4 −2 2 4 6 8 10
y [m]
Holes Antennas
Andrew Markham SenSys 2010 32/42
Motivation Steps in MI localization System Design Results Limitations and Lessons learnt Summary
Collars lasted for between 36 and 68 days Over 7 million MI readings delivered to basestation All generated data correctly received Average compression factor 21 Horizontal accuracy 45 cm Vertical accuracy 90 cm
Andrew Markham SenSys 2010 33/42
Motivation Steps in MI localization System Design Results Limitations and Lessons learnt Summary
5 6 7 8 9 10 11 x [m] 2 1 1 2 3 4 y [m]
2010-02-17 08:00:00 Badger C Badger D Hole
Andrew Markham SenSys 2010 34/42
Motivation Steps in MI localization System Design Results Limitations and Lessons learnt Summary
5 6 7 8 9 10 11 x [m] 2 1 1 2 3 4 y [m]
2010-02-17 09:00:00 Badger C Badger D Hole
Andrew Markham SenSys 2010 34/42
Motivation Steps in MI localization System Design Results Limitations and Lessons learnt Summary
5 6 7 8 9 10 11 x [m] 2 1 1 2 3 4 y [m]
2010-02-17 10:00:00 Badger C Badger D Hole
Andrew Markham SenSys 2010 34/42
Motivation Steps in MI localization System Design Results Limitations and Lessons learnt Summary
Position information of this resolution has never before been captured on an underground animal
Andrew Markham SenSys 2010 34/42
Motivation Steps in MI localization System Design Results Limitations and Lessons learnt Summary
1 3 5 7 9 11 x [m] 5 4 3 2 1
y [m]
Hole
Andrew Markham SenSys 2010 35/42
Motivation Steps in MI localization System Design Results Limitations and Lessons learnt Summary
Tracking system also reveals tunnel structure and resource usage
Andrew Markham SenSys 2010 35/42
Motivation Steps in MI localization System Design Results Limitations and Lessons learnt Summary
−80 −60 −40 −20 20 40 60
x [m]
−30 −20 −10 10 20 30 40 50
y [m]
Surface nodes Holes Detected
2010-02-19 01:00
Andrew Markham SenSys 2010 36/42
Motivation Steps in MI localization System Design Results Limitations and Lessons learnt Summary
−80 −60 −40 −20 20 40 60
x [m]
−30 −20 −10 10 20 30 40 50
y [m]
Surface nodes Holes Detected
2010-02-19 02:00
Andrew Markham SenSys 2010 36/42
Motivation Steps in MI localization System Design Results Limitations and Lessons learnt Summary
−80 −60 −40 −20 20 40 60
x [m]
−30 −20 −10 10 20 30 40 50
y [m]
Surface nodes Holes Detected
2010-02-19 03:00
Andrew Markham SenSys 2010 36/42
Motivation Steps in MI localization System Design Results Limitations and Lessons learnt Summary
−80 −60 −40 −20 20 40 60
x [m]
−30 −20 −10 10 20 30 40 50
y [m]
Surface nodes Holes Detected
2010-02-19 04:00
Andrew Markham SenSys 2010 36/42
Motivation Steps in MI localization System Design Results Limitations and Lessons learnt Summary
−80 −60 −40 −20 20 40 60
x [m]
−30 −20 −10 10 20 30 40 50
y [m]
Surface nodes Holes Detected
2010-02-19 04:00
Andrew Markham SenSys 2010 36/42
Motivation Steps in MI localization System Design Results Limitations and Lessons learnt Summary
5 6 7 8 9 10 11 12
x [m]
−2 −1 1 2 3 4 5
y [m]
Underground Trajectory Entrance Hole
2010-02-19 04:00
Andrew Markham SenSys 2010 36/42
Motivation Steps in MI localization System Design Results Limitations and Lessons learnt Summary
Seamless tracking between above and below ground components
Andrew Markham SenSys 2010 36/42
Motivation Steps in MI localization System Design Results Limitations and Lessons learnt Summary
Antenna resonance problems - works in the lab Premature collar failure - battery chemistry Limited tracking volume of 5 m x 5 m - easy to scale to larger areas Badgers playing with equipment - put in trees Robust hardware and software essential - only two chances per year, no reboot! micro-SD cards are tiny - attach tag to find when dropped
Andrew Markham SenSys 2010 37/42
Motivation Steps in MI localization System Design Results Limitations and Lessons learnt Summary
Optimal placement and sizing of antennas Increased resolution Large scale tests Application to other domains (aquatic, industrial) Integration with inertial and other sensors
Andrew Markham SenSys 2010 38/42
Motivation Steps in MI localization System Design Results Limitations and Lessons learnt Summary
First demonstration of MI as a feasible tracking technology for underground animals Allowed tracking which has never before been possible Real world deployment revealed interesting animal behaviour Extremely low power localization technique
Andrew Markham SenSys 2010 39/42
Motivation Steps in MI localization System Design Results Limitations and Lessons learnt Summary
The WildSensing Team: Bence, Cecilia, Kharsim, Rick, Salvo and Vlad Dr Chris Newman, Dr Christina Beusching and the Badger Project EPSRC (EP/E013678/1) and the Peoples Trust for Endangered Species
Andrew Markham SenSys 2010 40/42
Motivation Steps in MI localization System Design Results Limitations and Lessons learnt Summary
Andrew Markham andrew.markham@comlab.ox.ac.uk www.comlab.ox.ac.uk/people/andrew.markham/
Andrew Markham SenSys 2010 41/42
Motivation Steps in MI localization System Design Results Limitations and Lessons learnt Summary
Andrew Markham SenSys 2010 42/42