Position and Velocity Tracking in Mobile Cellular Networks Using the Particle Filter

1
Position and Velocity Tracking in MobileCellular
Networks Using the Particle Filter

• SourceIEEE WCNC
• AutherMohammed M. Olama
• Seddik M. Djouadi
• Chris S. Pendley
• Date2006/05/03
• SpeakerKuan-Ming Li

2
Outline

• Introduction
• Aulins Scattering Model
• State and Measurement Models
• Bayesian Bootstrap Filter Design
• Simulation results
• Conclusion

3
Introduction

• Particle Filter
• estimate the dynamic state (a set of random
  samples) of a moving object in a discrete time
• Bayesian methods is to construct a probability
  density function (PDF) of the state based on all
  the available information

4
Aulins Scattering Model(1/2)
1.
???
2.
Doppler shift ? velocity
3.
Phase shift ? location
5
Aulins Scattering Model(2/2)
6
State and Measurement Models

• System model
• xk fk(xk-1, wk-1)
• Measurement model (observation equation)
• zk hk(xk, vk)
• ktime step
• f(.,.) h(.,.) vector function
• wk vk white noise

7
Bayesian Bootstrap Filter Design(1/2)

• Assume that the initial PDF p(x0Z0) of the state
  vector
• At time step k, available information is the set
  of measurements Zk zii1,,k
• Need to construct the PDF of the current state xk
• A set of random samples xk(i) i1,,N draw
  from the PDF p(xkZk),N is the number of particle
• Update random sample xk1(i) i1,,N which are
  approximated as p(xk1Zk1)

8
Bayesian Bootstrap Filter Design(2/2)

• each sample is passed through the system model to
  obtain samples from the prior at time step k
• xk1(i) fk(xk(i), wk(i))
• With the measurement Zk, evaluate a normalized
  weight for each sample
• Resample N times from the discrete distribution
  to generate samples xk1(i) i1,,N, so that
  for any j, Prxk(j)xk(i) qi

9
Simulation results(1/3)

• Number of particles is 5000
• Number of time steps is 50
• The cell radius is 5000 meters
• , and are known
• The base station is located at the center of the
  cell and it is considered as the origin
• 2000 Hz for simulation reason
• RMSE(k)

10
Simulation results(2/3)
11
Simulation results(3/3)
12
Conclusions

• A new estimation mothod
• The result show that it is highly accurate and
  consistent