You are on page 1of 12

Progress Report 3

Experiment Set up
• Sixty finger print points separated by 3 ft
• Transmitter: HP signal generator
• Receivers: H600 spectrum analyzers
• Antenna: ??
• Transmitter power: 14.0 dBm
• Transmission frequency: 463.0 MHz
Experiment Set up

Durham Hall 4th Floor North Wing


Measurement
• Measurement taken in two rounds
• First round measurements are used to
estimate propagation model
• Second round measurement is used to
measure performance of positioning
algorithm.
Positioning using propagation
Model
Approach:
- The optimal log-normal propagation model is determined
using least-square algorithm from the measured data.

- This model is then used to estimate separation distance


between a transmitter and a receiver from RSL measurement.

- The estimated distance is used in trilateration algorithm to


determine the position of the transmitter.
Positioning using propagation
Model (Cont.)
Model: Prec i = Pref i – 10nilog(di/do) i = 1,2,3,4

Prec: Measured
Pref i : Estimated for each receiver from measured data
ni: Estimated for each receiver from measured data

2
  Prec1  Pref 1  
 10 n1 
( x1  xm )  ( y1  ym )  10
2 2

  (xi,yi): Location of receivers
  Prec 2  Pref 2  
2
(xm,ym): Location of mobile transmitter
 10 n2 
( x2  xm )  ( y2  ym )  10
2 2

 

  Prec 3  Pref 3  


2 Optimal value of each Pref i and ni is
 
determined from measured data
10 n3
( x3  xm )  ( y3  ym )  10
2 2

 
2
  Prec 4  Pref 4  

( x4  xm )  ( y4  ym )  10
2 2 10 n4 

Two unknowns and four equations
 
Results
Estimated propagation model for RX1
PL model for Rx1
-10
Actual
Propagation Model
-20

-30

-40
RSL [dBm]

-50

-60

-70

-80
0 1 2
10 10 10
Distance [ft]
Results (Cont.)
Estimated propagation model for RX2
PL model for Rx2
-10
Actual
Propagation Model
-20

-30

-40
RSL [dBm]

-50

-60

-70

-80
0 1 2
10 10 10
Distance [ft]
Results (Cont.)
Estimated propagation model for RX3
PL model for Rx3
-20
Actual
Propagation Model
-30

-40
RSL [dBm]

-50

-60

-70

-80
0 1 2
10 10 10
Distance [ft]
Results (Cont.)
Estimated propagation model for RX4
PL model for Rx4
0
Actual
-10 Propagation Model

-20

-30
RSL [dBm]

-40

-50

-60

-70

-80
0 1 2
10 10 10
Distance [ft]
Results (Cont.)
Positioning Performance
Radius Fingerprinting Method Propagation Model Method

Two Receivers Four Receivers Two Receivers Four Receivers

3 ft 25.0 % 51.7 % 3.3 % 3.3 %

6 ft 53.3 % 80.0 % 5.0 % 6.7 %

9 ft 68.3 % 90.0 % 15.0 % 15.0 %

12 ft 80.0 % 98.3 % 20.0 % 33.3 %

15 ft 90.0 % 100 % 28.3 % 46.7 %

18 ft 90.0 % 100 % 38.3 % 61.7 %


Conclusions
Measured data does not fit to the optimal propagation
model very well.
(Deviation up to 10 dB)

This is possibly because of the NLOS condition which is


prevalent in the environment.

This resulted in poor positioning performance. This is


reflected as:
- Inaccurate estimation of distance between
transmitter and receiver using propagation model
- When distance estimation is inaccurate, the
trilateration algorithm does not converge in most cases.

You might also like