Theory of the Gaussian channel

Hai all,

When we say that a channel is gaussian, does it mean that it's impulse response is gaussian..?? Or does it mean that the noise in the channel has a gaussian distribution..? Am a little confused

So if I were to simulate a gaussian channel in matlab, what is the impulse response, h, that I should be using??

thanks in advance

matt

Re: Gaussian Channel

Defn of gaussian channel is that there is no frequency selectivity, or fading(channel gain variation with time). only the AWGN, as it name implies,additive white gaussian noise is present.

model

y=x+n;

if binary used:

x=sign(1-2*rand(1000,1));

n=randn(1000,1);

y=x+n;


For frequency selective channels:

channel will provide filtering effect of x, and there will be still AWGN

Gaussian Channel

It refers to noise present in the channel has Gaussion distribution.

Gaussian Channel

As far as I am concern, the impulse response depends on the transfer function of the channel and it is deterministic, it has noting do do with noise in the channel that is random. If you try to have a impulse response considering noise you will come across a stochastic system and believe me it is far from complicated to modelling.

If you have an AWGN channel, there is in the AWGN Channel addition of noise to the signal in amplitudes that is gaussian distributed .

Re: Gaussian Channel

As far as I am concern, the impulse response depends on the transfer function of the channel and it is deterministic, it has noting do do with noise in the channel that is random. If you try to have a impulse response considering noise you will come across a stochastic system and believe me it is far from complicated to modelling.

Click to expand...

Agreed that trying to incorporate the noise also into the system impulse response would make it far too complicated as it would no longer be a deterministic system, but does a gaussian channel not introduce any fading or frequency selective response in addition to the AWGN...Janath's reply seems to indicate that a gaussian channel withous the noise (without which i guess it would no longer be gaussian..:!::! would give the same bits at the output......

ie,

y = x ....:!::!:
Have I construed it correctly..??...

Is it a similar thing with a rayleigh channel.?? I mean does it also not have a frequency selective response but just distort the signal with a noise which now has a rayleig pdf..??:!:

- matt

Gaussian Channel

I can't see the expressions there are "!" marks on them.

I think you are confuse with the definition of gaussian channel. It has to have noise with a gaussian pdf distribuition, otherwise it is not a gaussian channel.

In order to simulate a gaussian channel or a rayleigh channel you must basicly perform the same procedure, which is add the noise content in the signal of interest.

See a block diagram in a communication sistem, you will have the signal, the block from the channel's transfer function and a adder that indicate that the signal is summed to noise in the channel.

I guess what Janat did was exactly this: Considered no distortion or attenuation in the frequency band of the channel : generated a ramdon digital signal named X and added noise to it "n" to find the output Y

Regards
Claudio