Prev: PDIST
Next: interactionplot maineffectsplot
From: Hosso Hosso on 2 Jan 2010 06:39
Dear Friends,
I am trying to simulate a fading channel using Matlab's rayleighchan() function, my question is how one can assure the generated fading is slow or fast. What i know is that the condition for slow fading is Tc >> Ts otherwise it is fast fading , where Tc is the coherence time and Ts is the symbol duartion.
I am using the function rayleighchan(Ts,Fd) so that the above equation is achieved it seems to work with slow fading but what about fast fading??
I think the function is always assuming slow fading. Is this right?
Thanks,
Hoseb
From: nanren888 on 2 Jan 2010 14:53
"Hosso Hosso" <gamavor_535(a)hotmail.com> wrote in message <hhnb8n$m34$1(a)fred.mathworks.com>...
> Dear Friends,
> I am trying to simulate a fading channel using Matlab's rayleighchan() function, my question is how one can assure the generated fading is slow or fast. What i know is that the condition for slow fading is Tc >> Ts otherwise it is fast fading , where Tc is the coherence time and Ts is the symbol duartion.
> I am using the function rayleighchan(Ts,Fd) so that the above equation is achieved it seems to work with slow fading but what about fast fading??
> I think the function is always assuming slow fading. Is this right?
> Thanks,
> Hoseb

I have not used this function, but maybe I can offer some small insight.
Not sure what you mean by "slow or fast".
Commonly, "fast" is used to refer to the Rayleigh, multipath scattering & "slow" is used to refer to the slower, shadowing, often modelled as log-normal.

The quick answer is that if "slow fading" means to you "shadowing", then I doubt that a function named after Rayleigh will cover that, being more commonly considered lognormal.

Your question suggests, though that with "fast & slow" you might be interested in the fade rate.
The help on this says,
With the above two syntaxes, a smaller fd (a few hertz to a fraction of a hertz) leads to slower variations, and a larger fd (a couple hundred hertz) to faster variations.

(Not sure that I buy the "syntaxes" word)
I do have issues with characterisation of the fading with fd, but it should be as they simply say. The fd as a measure of the maximum Doppler should allow control of the fade rate.
If you have a spatial model in mind, you can picture a uniform Jakes scenario with consequent well defined max abs Doppler & that the fade rate seen will relate to this.
If you are simply driven from statistical signal processing, then I guess you imagine that the correlation & psd are transform pairs. The fading rate & hence correlation (in the absense of that nasty shadowing) are therefore related to what can be considered the Doppler frequency spectrum.
If you have simple assumptions on the distribution, eg uniform angle, or such, you can parameterise.
If that didn't help, perhaps vary fd & look for the effect.
From: Hosso Hosso on 2 Jan 2010 16:43
I will Check out what can i do, thank you for your help.


"nanren888 " <nanren888.remove.this(a)hotmail.com> wrote in message <hho86v$c32$1(a)fred.mathworks.com>...
> "Hosso Hosso" <gamavor_535(a)hotmail.com> wrote in message <hhnb8n$m34$1(a)fred.mathworks.com>...
> > Dear Friends,
> > I am trying to simulate a fading channel using Matlab's rayleighchan() function, my question is how one can assure the generated fading is slow or fast. What i know is that the condition for slow fading is Tc >> Ts otherwise it is fast fading , where Tc is the coherence time and Ts is the symbol duartion.
> > I am using the function rayleighchan(Ts,Fd) so that the above equation is achieved it seems to work with slow fading but what about fast fading??
> > I think the function is always assuming slow fading. Is this right?
> > Thanks,
> > Hoseb
>
> I have not used this function, but maybe I can offer some small insight.
> Not sure what you mean by "slow or fast".
> Commonly, "fast" is used to refer to the Rayleigh, multipath scattering & "slow" is used to refer to the slower, shadowing, often modelled as log-normal.
>
> The quick answer is that if "slow fading" means to you "shadowing", then I doubt that a function named after Rayleigh will cover that, being more commonly considered lognormal.
>
> Your question suggests, though that with "fast & slow" you might be interested in the fade rate.
> The help on this says,
> With the above two syntaxes, a smaller fd (a few hertz to a fraction of a hertz) leads to slower variations, and a larger fd (a couple hundred hertz) to faster variations.
>
> (Not sure that I buy the "syntaxes" word)
> I do have issues with characterisation of the fading with fd, but it should be as they simply say. The fd as a measure of the maximum Doppler should allow control of the fade rate.
> If you have a spatial model in mind, you can picture a uniform Jakes scenario with consequent well defined max abs Doppler & that the fade rate seen will relate to this.
> If you are simply driven from statistical signal processing, then I guess you imagine that the correlation & psd are transform pairs. The fading rate & hence correlation (in the absense of that nasty shadowing) are therefore related to what can be considered the Doppler frequency spectrum.
> If you have simple assumptions on the distribution, eg uniform angle, or such, you can parameterise.
> If that didn't help, perhaps vary fd & look for the effect.
From: Ethem Sozer on 4 Jan 2010 10:05
I want to point out that, Ts in "Tc >> Ts" is for symbol duration. But the
Ts in "rayleighchan(Ts,fd)" is for sampling time (period). So, while using
the MATLAB function, the symbol duration, let's call it Tsym, can be written
as:

Tsym = Ts * Nsamp;

where Nsamp is number of samples per symbol. If you use 10 samples to
represent a symbol, and your sampling time is 0.1 msec/sample, then your
symbol duration is 1 msec.

Hth,
Ethem


"Hosso Hosso" <gamavor_535(a)hotmail.com> wrote in message
news:hhoel7$k46$1(a)fred.mathworks.com...
>I will Check out what can i do, thank you for your help.
>
>
> "nanren888 " <nanren888.remove.this(a)hotmail.com> wrote in message
> <hho86v$c32$1(a)fred.mathworks.com>...
>> "Hosso Hosso" <gamavor_535(a)hotmail.com> wrote in message
>> <hhnb8n$m34$1(a)fred.mathworks.com>...
>> > Dear Friends,
>> > I am trying to simulate a fading channel using Matlab's rayleighchan()
>> > function, my question is how one can assure the generated fading is
>> > slow or fast. What i know is that the condition for slow fading is Tc
>> > >> Ts otherwise it is fast fading , where Tc is the coherence time and
>> > Ts is the symbol duartion.
>> > I am using the function rayleighchan(Ts,Fd) so that the above equation
>> > is achieved it seems to work with slow fading but what about fast
>> > fading??
>> > I think the function is always assuming slow fading. Is this right?
>> > Thanks,
>> > Hoseb
>>
>> I have not used this function, but maybe I can offer some small insight.
>> Not sure what you mean by "slow or fast".
>> Commonly, "fast" is used to refer to the Rayleigh, multipath scattering &
>> "slow" is used to refer to the slower, shadowing, often modelled as
>> log-normal.
>>
>> The quick answer is that if "slow fading" means to you "shadowing", then
>> I doubt that a function named after Rayleigh will cover that, being more
>> commonly considered lognormal.
>>
>> Your question suggests, though that with "fast & slow" you might be
>> interested in the fade rate.
>> The help on this says,
>> With the above two syntaxes, a smaller fd (a few hertz to a fraction of a
>> hertz) leads to slower variations, and a larger fd (a couple hundred
>> hertz) to faster variations.
>>
>> (Not sure that I buy the "syntaxes" word)
>> I do have issues with characterisation of the fading with fd, but it
>> should be as they simply say. The fd as a measure of the maximum Doppler
>> should allow control of the fade rate. If you have a spatial model in
>> mind, you can picture a uniform Jakes scenario with consequent well
>> defined max abs Doppler & that the fade rate seen will relate to this.
>> If you are simply driven from statistical signal processing, then I guess
>> you imagine that the correlation & psd are transform pairs. The fading
>> rate & hence correlation (in the absense of that nasty shadowing) are
>> therefore related to what can be considered the Doppler frequency
>> spectrum.
>> If you have simple assumptions on the distribution, eg uniform angle, or
>> such, you can parameterise.
>> If that didn't help, perhaps vary fd & look for the effect.


 | 
Pages: 1
Prev: PDIST
Next: interactionplot maineffectsplot