A Survey on Tracking Mobile Objects in Wireless Sensor Networks

1
A Survey on Tracking Mobile Objects in Wireless
Sensor Networks

• Presenter Bibudh Lahiri
• Course CPRE 546X
• Instructor Dr. Daji Qiao

2
Organization

• Problem statement
• Different approaches
• Tracking of mobile nodes
• Solution overview
• Challenges/issues

3
Problem statement

• Tracking the location of a mobile object in a 2D
  WSN
• Applications
• Surveillance of enemy soldiers
• or tanks in a battlefield
• Tracking habitat movement in jungle
• Health monitoring of patients/seniors

4
Different approaches

• Regional matching
• Hierarchical clustering/partition
• Prediction-based reporting
• Geometric, using binary proximity sensors
• Entropy-based, information-theoretic
• Differential game-theoretic
• Pointer-based distributed directory protocols

5
Different approaches Regional Matching

• A Regional Matching is a collection of pairs of
  read-write sets RW Read(v), Write(v) v e V
• RW is an m-regional matching if Write(v) n
  Read(u) ? F for all v,u e V such that dist(u,v)
  lt m
• When v tracks a user, it reports all vertices in
  Writei(v)
• When w looks for a user, it queries all vertices
  in Readi(w)
• 2i-regional matching ensures that dist(v,w) lt 2i
  implies Writei(v) n Readi(w) ? F

6
Different approaches Regional Matching

• Updating pointers near the user is relatively
  cheap
• Pointers near the user required to be more
  accurate
• Distant pointers updated less often

7
Different approaches Hierarchical Clustering
8
Different approaches Hierarchical Clustering

• STALK (Demirbas et al)
• find a mobile object at distance d
• needed O(d) time and communication
• move an object to distance d needed O(d log D)
  time and communication
• Too complex to implement on real motes

9
Different approaches Hierarchical Clustering

• Dynamic clustering for acoustic target
  tracking (Chen et al)
• Cluster formed and clusterhead became active,
  when signal strength exceeded a threshold
• Only one active clusterhead at a time
• No more than enough number of sensors replied

10
Different approaches Hierarchical Clustering

• LOCI
• clusters of bounded
• radius R, mR
• Implemented on MICA motes

11
Different approaches Prediction-based Reporting

• Sensor node having mobile object predicts
  objects movement for next reporting period
• Base station makes the same prediction based on
  the same object's movement history
• If observed movement matches nodes prediction,
  no transmission is needed
• Otherwise, sensor node corrects base station

12
Different approaches Binary Proximity Sensors

• Binary sensor provides 1-bit data about targets
  presence or absence
• Trajectory during a small interval approximated
  by straight line

13
Different approaches Binary Proximity Sensors

• Worst-case deviation between the estimated and
  the actual paths is at least O(1/?R)
• For fixed R, accuracy improves linearly with
  increasing ?
• For fixed number of sensors, accuracy improves
  linearly with increase in R

14
Different approaches Information Theory

• Entropy-based sensor selection heuristic
• Sensors observe target to reduce the uncertainty
  about target state
• Selective use of informative sensors needed less
  sensors to obtain the target state
• Prolonged system lifetime

15
Different approaches Game Theory

• Pursuers try to protect a linear target by
  intercepting the evaders as far as possible from
  the target
• Minimize the time to capture all evaders

16
Tracking of mobile nodes

• First discussed by Balakrishnan et al

17
Tracking of mobile nodes (contd.)

• Active mobile
• Mobile device actively transmits BEACONs
• Infrastructure nodes at known locations estimate
  distances based on received BEACONs
• Each receiver propagates this distance to a
  central database
• Multiple receivers concurrently obtain distance
  estimates to mobile device
• Simultaneity of distance samples guaranteed

18
Tracking of mobile nodes (contd.)

• Passive mobile
• Infrastructure nodes send BEACONs to passively
  listening mobile device
• Mobile device estimates distances to the
  infrastructure nodes
• Scales better as channel use is independent of
  number of mobile devices
• Receiver hears only one BEACON at a time, and may
  move between chirps from different infrastructure
  nodes
• Simultaneity of distance samples not guaranteed

19
Solution overview The Arrow protocol
20
Solution overview (contd.)

• Frame format

corresponding_beacon_seq_no
src
dst
seq_no
frame_type
8 bits
8 bits
8 bits
8 bits
8 bits
21
Solution overview (contd.)

• Active mobile
• Mobile node pro-actively and periodically
  broadcasts BEACONs
• Static node receiving the BEACON detects the
  mobile node
• Static node detecting the object sends RELAY
  frame to its neighbor
• Neighbor points to the sender of RELAY, and
  forwards RELAY to its neighbor

22
Solution overview (contd.)

• Passive mobile
• Static trackers periodically broadcast PROBE
  frames to see if object is near
• PROBE carries identifier of sender
• Upon receiving PROBE, mobile node unicasts BEACON
  frame to the sender of PROBE
• Targeted receiver of BEACON sends RELAY frame to
  its neighbor

23
Challenges/Issues

• Active mobile or passive mobile?
• Multiple instances of detection in active
  mobile approach

24
Challenges/Issues (contd.)

• To deal with multiple instances of detection
• Unicast BEACONs Add a destination address to
  BEACONs
• In TOSSIM, limit broadcast range of BEACON by
  adjusting bit error rates

25
Challenges/Issues (contd.)

• Concurrency
• In passive mobile approach, mobile node can
  receive new PROBE before its response to previous
  probe is spread across the network

26

• Thank You