Transcript Simulation Model

Simulation Model for
Mobile Radio Channels
Ciprian Romeo Comşa
Iolanda Alecsandrescu
Andrei Maiorescu
Ion Bogdan
[email protected]
Technical University “Gh. Asachi” Iaşi
Department of Telecommunications
Radio channel
Radio channel: propagation medium characterized by wave phenomena.
July, 2002
Slide 2
Technical University “Gh. Asachi” Iaşi
Department of Telecommunications
2
Fading
The propagation is realized mostly by reflection and diffraction.
Waves are received on different propagation ways => Multi-path Propagation.
The sum of waves received may have significant variations even on slow motion of receiver.
This is called short-term fading or fast fading and follows a Rayleigh distribution.
July, 2002
Slide 3
Technical University “Gh. Asachi” Iaşi
Department of Telecommunications
3
Fading
The propagation is realized mostly by reflection and diffraction.
Waves are received on different propagation ways => Multi-path Propagation.
The sum of waves received may have significant variations even on slow motion of receiver.
This is called short-term fading or fast fading and follows a Rayleigh distribution.
The mean of the received signal has slow variations on larger motion.
This is called long-term fading and follows a log-normal distribution.
July, 2002
Slide 4
Technical University “Gh. Asachi” Iaşi
Department of Telecommunications
4
Channel Modeling


A channel model has to allow the evaluation of the propagation loses and theirs
variations (fading).
The Suzuki model takes into account short-term fading with superimposed longterm log-normal variations of the mean of received signal:
 (t )   (t )   (t )
July, 2002
Slide 5
Technical University “Gh. Asachi” Iaşi
Department of Telecommunications
5
Analytical model – Stochastic process:
 (t )

 (t )  
Extended Suzuki Stochastic process


(t )
Log-normal process
models the short-time fading
is obtained considering:
complex zero mean Gaussian noise process  (t )  1 (t )  j  2 (t )
with cross-correlated quadrature components 1 (t ) and 2 (t )
 LOS component supposed to be independent of time (for short-time fading)

m  m1  j  m2    e
j  

is obtained as envelope of nonzero mean Gaussian noise process
  (t )   (t )  m
 (t )  (1 (t )  m1 )2  ( 2 (t )  m2 )2

For particular values of environment parameters, this process
follows Rice, Rayleigh or one-sided Gaussian distribution.
July, 2002
Slide 6
Technical University “Gh. Asachi” Iaşi
Department of Telecommunications
6
Analytical model – Log-normal process:
 (t )

 (t )  
Extended Suzuki Stochastic process


(t )
Log-normal process
models the long-time fading, caused by shadowing effects
is obtained from another real Gaussian noise process 3 (t )
with zero mean and unit variance:
 (t )  em s (t )
3


July, 2002
m and s are two environment parameters
3 (t ) and  (t ) are uncorrelated
Slide 7
Technical University “Gh. Asachi” Iaşi
Department of Telecommunications
7
Simulation Model


1 (t ), 2 (t ) cross-correlated
Simulation coefficients:
 c
-Doppler coefficients
i ,n
 f
-discrete Doppler
i ,n
frequencies
 
- Doppler phases
i ,n

N1 , N 2 = number of
sinusoids used to
approximate the
Gaussian processes
July, 2002
Slide 8
Technical University “Gh. Asachi” Iaşi
Department of Telecommunications
8
Simulation
A mixed signal simulation tool is used – Saber Designer with MAST language
MAST = HDL => the channel model can be used for simulations deeper to
hardware systems


Simulation Data
Environment parameters:






Number of sinusoids: N1=25 and N2=15.
Number of samples NS=108 and sampling period Ta=3·10-8s.
Maximum Doppler frequency fmax=91Hz, corresponding to a vehicle’s speed of
110Km/h.
Doppler coefficients ci,n and discrete Doppler frequencies fi,n are calculated at
the beginning and kept constants during the simulation.
Doppler phases θi,n are modified at each simulation step given by sampling.
July, 2002
Slide 9
Technical University “Gh. Asachi” Iaşi
Department of Telecommunications
9
Simulation results (1)
Envelope of the simulated extended Suzuki process
July, 2002
Slide 10
Technical University “Gh. Asachi” Iaşi
Department of Telecommunications
10
Simulation results (2)


The differences between the generated signal distribution obtained as
histogram and the analytical pdf are hardly observable.
The values for mean and standard deviation confirms this affirmation.
July, 2002
Slide 11
Technical University “Gh. Asachi” Iaşi
Department of Telecommunications
11
Conclusion
Histogram of simulated extended Suzuki model, in cases of:

Light shadowing  log-normal distribution

Heavy shadowing  Rice (or Rayleigh) distribution
July, 2002
Slide 12
Technical University “Gh. Asachi” Iaşi
Department of Telecommunications
12