| Author |
Message |
urwelcome
Joined: 12 Apr 2007
Posts: 86
Helped: 6
|
 14 Jul 2008 13:18 channel capacity bpsk |
|
|
|
|
capacity formula
C=log(1+SNR)
is valid when input symbols have gussian distribution (please correct me if I am wrong)
but when input symbols are not gaussian distributed but are uniformly distributed binary symbols than how to find out the capacity for it ??
please recommend me some references ..
|
|
| Back to top |
|
 |
Google
AdSense
|
14 Jul 2008 13:18 Ads |
|
|
|
|
|
|
| Back to top |
|
|
aliazmat
Joined: 14 Mar 2008
Posts: 77
Helped: 3
|
17 Jul 2008 11:15 channel capacity for bpsk |
|
|
|
|
| You are talking about Discrete Input and Contiuous Output channel, go through Proakis chapter 7.
|
|
| Back to top |
|
|
senaydud
Joined: 23 Jun 2008
Posts: 150
Helped: 28
|
18 Jul 2008 1:05 bpsk capacity |
|
|
|
|
| Hi urwelcome, To be honest I dont know the exact answer for your question but to me the formula C=B*log(1+SNR) gives only an upper limit. It is not directly related with Gaussian. Yes it is calculated by assuming a gaussian distribution and I hope you somehow can calculate it for other distributions but in fact it says: the highest Rate that you can achieve using a communication system is C=B*log(1+SNR) whether it is gaussian distributed or not.
|
|
| Back to top |
|
|
bulx
Joined: 07 Aug 2004
Posts: 171
Helped: 24
|
18 Jul 2008 12:11 bpsk channel capacity |
|
|
|
|
May be this constitutes an answer:
Like you have said, capacity is achieved when input to the AWGN channel is gaussian distributed. Thus, when input in not gaussian, capacity is NOT achieved. So the question on what is the capacity when input is equi-probable binary is incorrect, AFAIK.
You can only say that with this binary signalling scheme, you chose to operate the channel with a rate lower than capacity.
-b
|
|
| Back to top |
|
|
