Variance of Gaussian Noise

Vladimir Vassilevsky  <antispam_bogus@hotmail.com> wrote:

> Any non-AWGN case can be converted to the equivalent Gaussian
> by means of some linear or non-linear operation with the
> corresponding increase in the channel capacity.

Suppose the non-AWGN case consists of impulse noise arriving
randomly (say, with Poisson statistics).  How would you
convert that to the AWGN case?

(I agree that the capacity is higher given the same noise power
for the impulse-noise case; I'm unclear on how one could do the 
conversion.)

Steve

Steve Pope wrote:
> Vladimir Vassilevsky  <antispam_bogus@hotmail.com> wrote:
> 
> 
>>Any non-AWGN case can be converted to the equivalent Gaussian
>>by means of some linear or non-linear operation with the
>>corresponding increase in the channel capacity.
> 
> 
> Suppose the non-AWGN case consists of impulse noise arriving
> randomly (say, with Poisson statistics).  How would you
> convert that to the AWGN case?

First, I will limit those pulses to the level of the useful signal, then
I will smear the residue by some sort of dispersive filter.


Vladimir Vassilevsky
DSP and Mixed Signal Design Consultant
http://www.abvolt.com

Vladimir Vassilevsky  <antispam_bogus@hotmail.com> wrote:

>Steve Pope wrote:

>> Suppose the non-AWGN case consists of impulse noise arriving
>> randomly (say, with Poisson statistics).  How would you
>> convert that to the AWGN case?

>First, I will limit those pulses to the level of the useful signal, then
>I will smear the residue by some sort of dispersive filter.

Sounds good.  I think one could argue that if you apply enough
pseudorandom interleaving, the impulse noise becomes gaussian,
using some sort of central-limit-theorem argument.

Steve

On Mar 26, 11:40&#4294967295;am, Vladimir Vassilevsky <antispam_bo...@hotmail.com>
wrote:
> sasuke wrote:
> > The question which I actually wanted to ask was, is it true that Gaussian
> > noise is the worse kind of noise to affect a communication channel? In
> > other words is the capacity of a communication channel least when the
> > channel is corrupted by an additive Gaussian noise??
>
> You are correct. For the given noise power, the AWGN corresponds to the
> minimum capacity of the communication channel.

Agree.

>Any non-AWGN case can be
> converted to the equivalent Gaussian by means of some linear or
> non-linear operation with the corresponding increase in the channel
> capacity.

Not really. You may be able to convert the channel to a gaussian one,
but you can not increase the channel capacity by the data processing
theorem in information theory.
>
> Vladimir Vassilevsky
> DSP and Mixed Signal Design Consultanthttp://www.abvolt.com