Location and tracking of mobile users in a wireless system

1
Location and tracking of mobile users in a
wireless system

• Szymon Fedor

2
Application structure

• Space modelling
• Signal prediction at each cube centre
• Reading signal and comparing the value with the
  prediction
• Estimation of the user location

3
Application structure

• Space modelling
• Signal prediction at each cube centre
• Reading signal and comparing the value with the
  prediction
• Estimation of the user location

4
Space modelling

• Building and equipment modelled by the perfect
  geometric forms

Building.xml
5
(No Transcript)
6
Space modelling

• Space divided into equally dimensioned cubes

7
Space modelling

• Cube centres are the nominal user locations

8
Application structure

• Space modelling
• Signal prediction at each cube centre
• Reading signal and comparing the value with the
  prediction
• Estimation of the user location

9
Signal prediction

• Use of ray tracing method
• Signal received is a sum of many rays

10
Signal prediction

• Propagation environment with many obstacles due
  to constant movement and instability
• The signal is not perfectly predictable around
  the user
• The signal is composed of two parts
  deterministic amplitude and random phase

The phase of the rays is uniformly distributed
between 0 and 2?
The signal is composed of the sum of i rays
The amplitude of each ray is predicted with the
ray tracing method
11
Signal prediction

• The theory and the tests prove that

has a Rayleigh
distribution
Rayleigh distribution function
Rayleigh distribution graph for ?2
12
Signal prediction

• At each cube centre and for all base stations we
  estimate the ? parameter
• We use the Rayleigh distribution property

13
Application structure

• Space modelling
• Signal prediction at each cube centre
• Reading signal and comparing the value with the
  prediction
• Estimation of the user location

14
Reading signal and comparing the value with the
prediction

• Application periodically collects the signal
  from all base stations
• For tests purpose I simulated the signal value
  with the ray tracing method
• For each base station, we calculate the
  probability of user location at each cube centre

15
Application structure

• Space modelling
• Signal prediction at each cube centre
• Reading signal and comparing the value with the
  prediction
• Estimation of the user location

16
Estimation of the user location

• Use of Bayesian method to create location
  probability grid
• At the beginning the application estimates user
  location in the whole building
• For the next n times it looks for the user just
  in the neighbourhood of the previous location

17
Bayesian method

• The probability of user location is uniformly
  distributed between the search space
• The likelihood of user location at one cube is a
  multiplication of the estimation of all base
  stations

mj - the location cube, which is the possible
mobile position
- measurement in the vicinity of the lth station
18
Bayesian method

• By including the Rayleigh distribution function
  we get

mj - the location cube, which is the possible
mobile position
- measurement in the vicinity of the lth station

• The Bayes theorem gives the probability of
  occupation of each cube by a user

• The user location is associated with the cube
  which has
• the highest probability

19
Test results

• All tests with the same building structure and 7
  base stations distributed in the building

• The signal send by the user was simulated with
  the ray tracing method

20
Precision in function of the base station quantity
Likelihood uniformly distributed with the value
0,028
21
Precision in function of the base station quantity
Cube 4 estimated with the likelihood 0.14
22
Precision in function of the base station quantity
Cube 4 estimated with the likelihood 0.12
23
Precision in function of the base station quantity
Cube 4 estimated with the likelihood 0.28
24
Precision in function of the base station quantity
Cube 4 estimated with the likelihood 0.29
25
Precision in function of the base station quantity
Cube 4 estimated with the likelihood 0.34
26
Precision in function of the base station quantity
Cube 4 estimated with the likelihood 0.44
27
Precision in function of the base station quantity
Cube 4 estimated with the likelihood 0.47
28
Test for the cube dimension 6.6 m
The view from the top of the building of the
users trip
29
Test for the cube dimension 6.6 m
Wrong user location estimation but in the
neighbourhood
30
Test for the cube dimension 6.6 m
Correct user location estimation
31
Test for the cube dimension 6.6 m
Correct user location estimation
32
Test for the cube dimension 6.6 m
Wrong user location estimation but in the
neighbourhood
33
Test for the cube dimension 6.6 m
Correct user location estimation
34
Test for the cube dimension 6.6 m
Wrong user location estimation but in the
neighbourhood
35
Test for the cube dimension 6.6 m
Correct user location estimation
36
Test for the cube dimension 6.6 m
Wrong user location estimation but in the
neighbourhood
37
Conclusion

• Possible improvements
• - add another estimation parameter
• - change the distribution of the signal
• Tests in the real environment

Thank you