Welcome to EDAboard.com

Welcome to our site! EDAboard.com is an international Electronic Discussion Forum focused on EDA software, circuits, schematics, books, theory, papers, asic, pld, 8051, DSP, Network, RF, Analog Design, PCB, Service Manuals... and a whole lot more! To participate you need to register. Registration is free. Click here to register now.

Channel Capacity in AWGN

Status
Not open for further replies.

gama

Newbie level 4
Joined
Dec 4, 2005
Messages
7
Helped
0
Reputation
0
Reaction score
0
Trophy points
1,281
Activity points
1,329
awgn channel capacity

Hi,

in AWGN, we have channel capacity equation:
C = (½)log(1+SNR),
but if we use QAM16,
then the SNR means Eb/No or Es/No or Eav/No?

and even more if we are in Linear-filtering channel,
e.g, h(t)=0.8δ(t)-0.48δ(t-T)+0.36δ(t-2T),
then how we change the C equation or not?

thx.
gama
 

Aasem_sadek

Junior Member level 3
Joined
Jan 2, 2006
Messages
26
Helped
2
Reputation
4
Reaction score
1
Trophy points
1,283
Location
Egypt,Cairo
Activity points
1,501
channel capacity qpsk

i guess the channel capacity = C=B log(1+S/N) not 1/2 log(1+s/n)
where S=Eb*Rb ;Rb is the bit rate
N=No*B ;where No is the noise power spectral density & B is the band width of channel
so i guess if you used the Es u shall use the Rs which = Rb/k ;where k=log (base2) M where K is the number of bits per symbol
& Es then is equal to Eb *Log(base 2) M which will get you back to Eb*Rb
note that this is valid for grey coding only ie: 1 bit error only allowed in the symbol
 

changfa

Advanced Member level 4
Joined
Dec 14, 2005
Messages
119
Helped
21
Reputation
42
Reaction score
6
Trophy points
1,298
Activity points
2,124
channel capacity awgn

The capacity C=B log(1+S/N) is measured in [bits/sec] while C=1/2 log(1+s/n) is in [bits/sec/Hz]. No difference.

Concerning the multipath situation, the capacity does change for sure. But, for your 16QAM, I do not have any idea. In general, however, you can transform the original channel to parallel channel (for example, OFDM technique) and get the capacity of this parallel channel as sum of individual channels.

Following reference might be helpful:
Digital Communication
by Edward A. Lee, David G. Messerschmitt
(I am sure it answers your first quesiton)
 

amihomo

Full Member level 5
Joined
Jan 9, 2007
Messages
284
Helped
66
Reputation
132
Reaction score
42
Trophy points
1,318
Location
PERSIAN Gulf
Activity points
3,101
awgn equation

No Comment!

In an M-ary system :
Es/N0 (dB)=Eb/N0(dB) +log2(k) ,where k=code rate×log2(M)

(when coding is not used ,code rate is equal to 1)

Es/N0(dB)=10log10(Tsymbol/Tsampling) + SNR(dB)

and finally : C=B log(1+SNR)

It's that simple!
:D:D:D
 

changfa

Advanced Member level 4
Joined
Dec 14, 2005
Messages
119
Helped
21
Reputation
42
Reaction score
6
Trophy points
1,298
Activity points
2,124
channel capacity vs snr

amihomo said:
No Comment!

In an M-ary system :
Es/N0 (dB)=Eb/N0(dB) +log2(k) ,where k=code rate×log2(M)

(when coding is not used ,code rate is equal to 1)

Es/N0(dB)=10log10(Tsymbol/Tsampling) + SNR(dB)

and finally : C=B log(1+SNR)

It's that simple!
:D:D:D

Sorry, but you ignore the informational meaning of C=B log(s+SNR).

Take Tsymbol=Tsampling and no channel coding for example, your equation goes unbounded when EbNo goes to infinity. This is not true since we know for example QPSK the capacity is limited to 2bits/sec/Hz no matter how large EbN0 (SNR) is.
 

    V

    Points: 2
    Helpful Answer Positive Rating

ahmedseu

Full Member level 3
Joined
Jul 31, 2006
Messages
183
Helped
71
Reputation
142
Reaction score
43
Trophy points
1,308
Location
Bahrain
Activity points
2,422
capacity awgn channel

Hi,

I think this thesis is one of the best in this field.

Regards
 
  • Like
Reactions: Xavipersonal

    Xavipersonal

    Points: 2
    Helpful Answer Positive Rating
    V

    Points: 2
    Helpful Answer Positive Rating

amihomo

Full Member level 5
Joined
Jan 9, 2007
Messages
284
Helped
66
Reputation
132
Reaction score
42
Trophy points
1,318
Location
PERSIAN Gulf
Activity points
3,101
capacity of awgn

changfa said:
amihomo said:
No Comment!

In an M-ary system :
1) Es/N0 (dB)=Eb/N0(dB) +log2(k) ,where k=code rate×log2(M)

(when coding is not used ,code rate is equal to 1)

2) Es/N0(dB)=10log10(Tsymbol/Tsampling) + SNR(dB) , for complex input signals

3) and finally : C=B log(1+SNR)

It's that simple!
:D:D:D

Sorry, but you ignore the informational meaning of C=B log(s+SNR).

Take Tsymbol=Tsampling and no channel coding for example, your equation goes unbounded when EbNo goes to infinity. This is not true since we know for example QPSK the capacity is limited to 2bits/sec/Hz no matter how large EbN0 (SNR) is.

Hi,
Obviously there is no doubt that equation (1) and (3) are correct , to prove equation (2) :

Es/N0(dB)=10log10(S×Tsymbol/(N/Bn))
=10log10( (Tsymbol×Fs).(S/N) )
=10log10(Tsymbol/Tsampling) + 10log10(S/N )
=10log10(Tsymbol/Tsampling) + SNR(dB)

where :
S=input signal power ,(Watts)
N=noise power ,(Watts)
Bn=noise Bandwidth ,(Hz)
Fs=sampling frequency ,(Hz)

(note that: Bn=Fs=1/Tsampling)

what you said about Eb/N0->∞ also holds if we let SNR->∞ in eq. (3) . and this not related to eq. (1) and (2)
 

changfa

Advanced Member level 4
Joined
Dec 14, 2005
Messages
119
Helped
21
Reputation
42
Reaction score
6
Trophy points
1,298
Activity points
2,124
channel capacity snr

amihomo said:
changfa said:
amihomo said:
No Comment!

In an M-ary system :
1) Es/N0 (dB)=Eb/N0(dB) +log2(k) ,where k=code rate×log2(M)

(when coding is not used ,code rate is equal to 1)

2) Es/N0(dB)=10log10(Tsymbol/Tsampling) + SNR(dB) , for complex input signals

3) and finally : C=B log(1+SNR)

It's that simple!
:D:D:D

Sorry, but you ignore the informational meaning of C=B log(s+SNR).

Take Tsymbol=Tsampling and no channel coding for example, your equation goes unbounded when EbNo goes to infinity. This is not true since we know for example QPSK the capacity is limited to 2bits/sec/Hz no matter how large EbN0 (SNR) is.

Hi,
Obviously there is no doubt that equation (1) and (3) are correct , to prove equation (2) :

Es/N0(dB)=10log10(S×Tsymbol/(N/Bn))
=10log10( (Tsymbol×Fs).(S/N) )
=10log10(Tsymbol/Tsampling) + 10log10(S/N )
=10log10(Tsymbol/Tsampling) + SNR(dB)

where :
S=input signal power ,(Watts)
N=noise power ,(Watts)
Bn=noise Bandwidth ,(Hz)
Fs=sampling frequency ,(Hz)

(note that: Bn=Fs=1/Tsampling)

what you said about Eb/N0->∞ also holds if we let SNR->∞ in eq. (3) . and this not related to eq. (1) and (2)

The problem is your logic. In your first post, it seems that you want prove (3) by (1) and (2), and this procedure does not hold for sure. For your second post, you want use (1) and (3) to prove (2), which itself has nothing to do with (3) at all.

The bottomline is: the famous C=B log(1+P/NB) bits/sec is defined for Gaussian channel and this value C is an intrinsic property of the channel. When we talk about a specific modulation scheme, we have to calculate its capacity curve (the above reference by some friend might help), which is definitely below the famous C~SNR curve.
 

    V

    Points: 2
    Helpful Answer Positive Rating

amihomo

Full Member level 5
Joined
Jan 9, 2007
Messages
284
Helped
66
Reputation
132
Reaction score
42
Trophy points
1,318
Location
PERSIAN Gulf
Activity points
3,101
capacity awgn

Hi Changfa,
I didn't mean to prove eq (3) by (1) and (2), i meant to show how Eb/N0 and Es/N0 are related to the SNR in eq (3).

it is evident that C is an intrinsic property of the channel.As you know, by definition, channel capacity (C) is the maximum data rate at which reliable transmission of information over the channel is possible.And according to shannon's limit it is an upper bound for the reliable data rate we can achieve in any modulation scheme, i.e. , At rates R<C reliable transmission of information over the channel is possible .

regards
 

    V

    Points: 2
    Helpful Answer Positive Rating

spweda

Full Member level 3
Joined
Jun 28, 2001
Messages
184
Helped
7
Reputation
14
Reaction score
2
Trophy points
1,298
Activity points
1,261
capacity snr

The channel capacity is independent to the modulation format. If 16 QAM is used, it is more efficient than QPSK and more close to the channel capacity.
 

    V

    Points: 2
    Helpful Answer Positive Rating
Status
Not open for further replies.

Part and Inventory Search

Welcome to EDABoard.com

Sponsor

Top