White noise in communication channel

[LM741]

hi all - its been very long - hope everyone is well!

just a data comms related question:

When modelling a communication channel we normally include AWGN, that is, Additive white gaussian noise.

Can anybody tell me why we use white noise in the model. I know it has a constant power spectral density for all frequencies - but how does this help us or make things better?
Also, apperently the DC component( mean) of the noise signal is zero - is this because we use the Gaussian distribution? i.e. how to we get it to equal zero?

thanks very much

John

 

[light_bulb]

sqrt(4kTBR)=Vn

short answer: the noise is proportional to resistance, temp and bandwidth. you can get it equal by using a balanced circuit where these variables are the same and have common mode rejection reduce it to dc.













is my guess

 

[marcusl]

Hello 741!

1) Thermal receiver noise is white, as mentioned above, and counts as part of the channel. Other noise, including that from motors, transmitters, lightning, etc. is random as well so, on average, at any time position and frequency, it's described well by Gaussian statistics.
2) DC is irrelevant to a high frequency communications system and isn't even picked up by the antenna or electronics. (Same is true of ethernet and other cabled digital channels.) Hence zero mean.


BTW, Did you figure out the sampled data/DFT stuff?

 

[LM741]

sup marcusl! thanks
ha ha - man can't believe you remember that stuff! yeah finaaly grabbed it by the horns and grasped it! actually ended up failing Signals and Systems 2 end of year - but there is a happy ending to the story - coz i qualified for a supp in January and passed - so thanks for all your help!
thanks ligh_bulb
so marcus - you telling me that the white noise is used to represent thermal noise and random amplidutes ( created by Gaussina) are used to repreent random bursts of noise - so in essence, we are represnting two types of noise? but what is so special about a CONSTANT power spectral density? thanks

 

[marcusl]

Good, glad it worked out!

Over large spans the noise power spectral density (PSD) can vary--it's different at 1MHz and 10GHz--but over any typical narrow channel it just doesn't vary much. For example, Verizon cell phones operate on a protocol called CDMA, occupying a 1.5 MHz channel located within one of Verizon's 5 MHz wide frequency allocations. Center frequency is 1.9GHz. Over this narrow channel a constant average PSD is a good approximation.

 

[LM741]

thanks
....