Sensor networks for traffic monitoring

1
Sensor networks for traffic monitoring
Pravin Varaiya et al
2
Outline

• Challenge
• Sensor networks for traffic applications
• Pedamacs MAC protocol
• Signal processing

3
Challenge

• Accuracy and low delay
• Biggest cost is deployment and maintenance-lifetim
  e (power consumption) will determine economic
  feasibility

4
Sensor networks for traffic
lt 100 m
Access point
Sensor node
Freeway
Intersection

• Nodes generate data, report to access point
• At intersection, vehicle detection must be
  reported in 0.1 s also 30-sec periodic data
• Nodes are power- and energy-limited access
  points are not

5
Current traffic monitoring technology

• Loop detectors is the standard loops last 10
  years
• Closing lane to cut loops in freeway pavement is
  very disruptive
• Alternatives today are microwave radar, video
  cameras
• Installed cost is 600-1000 per detector (lane)
  per year
• Can sensor networks compete?

6
Sensor networks with two special characteristics

• One distinguished node , Access Point or AP
  sensor nodes or SN periodically (eg. 30 s)
  generate data for transmission to access point
• SNs are power- and energy-limited but AP is not
  Consequently
• Transmission AP ? SN is one-hop
• Transmission SN ? AP is multi-hop
• Two conditions satisfied in traffic applications

Freeway
7
Pedamacs vs random access networks

• Pedamacs networks
• Access point discovers network topology nodes
  discover next hop
• Access point computes and broadcasts transmission
  schedule to all nodes (TDMA data)
• During data phase, node sleeps if it is not
  scheduled to listen or to transmit

• Random access networks
• Access point and nodes discover next hop
• Nodes randomly transmit and constantly listen for
  incoming packets
• Refinements proposed to reduce node listening
  time

8
Comparison of random access and Pedamacs networks

• Comparison via TOSSIM, a TinyOS simulator
• Need to select critical parameters for comparison
• Backoff-listening random access scheme
• Back-off window, listening window
• Transmission range
• Nodes randomly distributed inside unit circle

9
Power consumption in PEDAMACS vs random access

• 50 kbps one packet every 30 sec vertical scale
  is log10
• Listening in random access uses 1000X more power,
  and receiving uses 10X power than in Pedamacs

10
Lifetime of PEDAMACS vs random access network

• Two AA batteries 2200 mA at 3 V
• Pedamacs network lasts 600 days, need 5X
  improvement
• Random access network lasts 10 days-unsuited for
  traffic control

11
Pedamacs vs random access delay

• Random access delay excessive for traffic
  application

12
Detecting vehicles

• Experiments
• Spatial and temporal resolution
• Speed
• Vehicle classification

13
Data Set 1 motes in middle of lane
1 Mic
3 MagXY
2 Mic
4 MagXY
14
Ford15_x0_1.dat
Wind disturbance
15
Ford27_x0_1_track_at_end.dat
Noise from truck
16
Ford15_x0_1.dat
Magnetic signature for classification
ford27_x0_1_track_at_end.dat
17
Ford25_x0_2.dat
18
Ford_acc_x0_1.dat
Vehicle accelerating going over the mote
19
ford_stopB4mote1_1sec_acc_x0_1_otherCars.dat
From another car
Car stopped before mote 1,3
20
Summary

• Sensor networks offer a promising alternative
• Acoustic signal is corrupted by noise--more
  filtering and processing needed for robust
  detection
• Magnetic signal depends on orientation
• Work needed to implement TDMA protocol
• Signal processing for speed, vehicle
  classification
• Deployment, reliability