magneto-inductive networked rescue system (miners): taking sensor networks underground

Magneto-Inductive NEtworked Rescue System (MINERS): Taking Sensor Networks Underground

Andrew Markham and Niki TrigoniDepartment of Computer ScienceUniversity of Oxford

>2500 large underground mines

Mining disasters

Collapse: Chile

Fire/Explosion: Sago, USA

Flooding: Wales

Existing approaches

Wired

Wired

Wired

! !

?

Wireless

Wireless

Wireless

! !

?

Need

Need

! !

Need

! !

Solution: Magneto-Inductive

Overview

• Motivation• System Requirements and Design• Challenges and Solutions• Implementation and Results• Summary

System Requirements

• Nodes should not require precise positioning

• Nodes should be able to communicate over 30m through rock

• Information must be relayed to surface with minimal latency

Magneto-Induction

Magneto-Induction

Magneto-Induction

V

Overview

• Motivation• System Requirements and Design• Challenges and Solutions• Implementation and Results• Summary

Challenges

• Placement• Path loss and noise• Limited bandwidth• Latency

Challenge: Placement

1.00 V

𝑉 ∝𝐵 cos𝜃

Challenge: Placement

0.70 V

𝑉 ∝𝐵 cos𝜃

Challenge: Placement

0.00 V

𝑉 ∝𝐵 cos𝜃

Solution: Triaxial Receivers

𝑉 ∝|𝐵|

1.00 V

Solution: Triaxial Receivers

𝑉 ∝|𝐵|

1.00 V

Solution: Triaxial Receivers

𝑉 ∝|𝐵|

1.00 V

Solution: Triaxial Receivers

𝑉 ∝|𝐵|

1.00 V

Rotationallyinvariant

Challenges

• Placement• Path loss and noise• Limited bandwidth• Latency

Challenge: Path loss and noise

Frequency (Hz)

Noi

se (d

B)

Background noise

Challenge: Path loss and noise

Frequency (Hz)

Atten

uatio

n (d

B)

Skin effect

Solution: VLF Narrowband

Frequency (Hz)Pow

er S

pect

ral D

ensi

ty

32 Hz

2500 Hz

Avoid LF noise Minimize skin effect

Solution: VLF Narrowband

Frequency (Hz)Pow

er S

pect

ral D

ensi

ty

32 Hz

2500 Hz

Avoid LF noise Minimize skin effect

Sweet Spot

Challenges

• Placement• Path loss and noise• Limited bandwidth• Latency

Challenge: Limited bitrate

32 Hz

BPSK: 32 bps!

Vector fields

Vector fields

Vector fields

Vector fields

=[2;0;0]

Vector fields

=[0;0;-1]

Vector fields

=[0;-1;0]

Electrically Rotating Transmitter

X = +1Amp

Sending symbol 0

Transmitter Receiver

Electrically Rotating Transmitter

Y = +1Amp

Sending symbol 1

Transmitter Receiver

Electrically Rotating Transmitter

Z = +1Amp

Transmitter Receiver

Sending symbol 2

Electrically Rotating Transmitter

X = -1Amp

Sending symbol 3

Transmitter Receiver

Solution: Magnetic Vector Modulation

BPSK:2 symbols: 32bps 01

Solution: Magnetic Vector Modulation

BPSK:2 symbols: 32bps

Magnetic vector modulation:6 symbols: 80bps

01

2

3

4

01

Solution: Magnetic Vector Modulation

BPSK:2 symbols: 32bps

Magnetic vector modulation:6 symbols: 80bps

~2.5 times increase inbitrate

Same energy

Solution: Magnetic Vector Modulation

BPSK:2 symbols: 32bps

Magnetic vector modulation:6 symbols: 80bps

01

2

3

4

01

Solution: Magnetic Vector Modulation

BPSK:2 symbols: 32bps

Magnetic vector modulation:6 symbols: 80bps

01

2

3

4

01

Solution: Magnetic Vector Modulation

BPSK:2 symbols: 32bps

Magnetic vector modulation:6 symbols: 80bps

01

2

3

4

01

Need to train channel

Challenges

• Placement• Path loss and noise• Limited bandwidth• Latency

Challenge: Latency

• Rapid query response essential• How many miners are trapped

underground?• What is the maximum methane

concentration?

Challenge: Latency

A B

Challenge: Latency

A B

Challenge: Latency

A B

Challenge: Latency

A B

2 units of time

Challenge: Latency

Depth

Breadth

T≈ Breadth x Depth

Narrow transmitter bandwidth

Frequency (Hz)Pow

er S

pect

ral D

ensi

ty

32 Hz

2500 Hz

Wider receiver bandwidth

Frequency (Hz)Pow

er S

pect

ral D

ensi

ty

1000 Hz

1 2 20

Solution: Broadcatching

A B

Solution: Broadcatching

A B

1 unit of time

Solution: Broadcatching

Depth

Breadth

T≈ Depth

Solution: Broadcatching

Depth

Breadth

T≈ Depth

Reduced latency

Overview

• Motivation• System Requirements and Design• Challenges and Solutions• Implementation and Results• Summary

Implementation: System

Implementation: System

Implementation: Miner unit

Implementation: Miner unit

Implementation: Transmitter

Implementation: Receiver

Message format

Message format

Phase lock

Message format

Training

Message format

Data

Received Message

Received Message

Strongest on y channel

Training: Transmit on X

Bx

By

Bz

0

3

Training: Transmit on Y

Bx

By

Bz

1

4

Training: Transmit on Z

Bx

By

Bz2

5

Symbol Constellation

Symbol Constellation

Symbol Constellation

Increase in data rate

Latency

Latency

Latency

Latency

5 fold decrease in latency

Real world trials

Real world trials: BER

Real world trials

• Communication through solid rock• 25m range at 80 bps• 75m range at 0.1 bps

Real world trials

• Communication through solid rock• 25m range at 80 bps• 75m range at 0.1 bps

MI is a viable technique

Future Directions

• ARM based SDR• More sensitive receiver• Mobile MI nodes• Testing in operational mines

Conclusion

• Need for robust communication in mining• Modulate vector field to increase bitrate• Broadcatching to reduce latency• Prototype SDR transceiver• First real world test of MI sensor network• Potential lifesaving technology

Thanks

• JP van de Ven• Oxford Martin School• EPSRC SUAAVE

Questions?

andrew.markham@cs.ox.ac.uk