No in MMSE Equalization

Status
Not open for further replies.
G

Ghalzai

Newbie level 3
Joined
Dec 30, 2007
Messages
4
Helped
0
Reputation
0
Reaction score
0
Trophy points
1,281
Activity points
1,310
Hi all, I am facing problem in MMSE Equalization. I don't know how to calculate my No (noise power spectral density).

I am generating noise using awgn. Can any one tell me how should i calculate my No???

Secondly If I generate noise using randn, then on comparison between ZF and MMSE, i am not getting good results for MMSE. I think i am not generating enough noise. Can any one tell me how can i increase my noise.
 

Use AWGN function and set SNR as its input to control amount of noise, or simply multiplr the randn output by a coefficient and the noise increases as this coeff. inc.. I think MMSE needs No as an input, otherwise you have to implement an SNR estimation algorithm in conjunction with your equalizer
 

Thanks for your valuable reply.
Can you please tell me if i use awgn function then how should i calculate No?

Actually i am using approch for MMSE which is discussed by Proakis. Here we have to calculate No and i dont know how to calculate it. ((
 

For awgn, you use the SNR in dB as input, the SNR is Eb/No-> Bit Energy / Noise PSD thus No = Eb / 10^(SNR/10)
 

so Eb is required to calculate it. How to calculate Eb?? :O
In my program i am using adaptive Modulation scheme. means depending on the channel snr, we use number of bits sent on that channel and then determine its modulation scheme. So my number of bits sending in symbol are varying from 2-7.
 

Hi Ghalzai,

I guess you are performing a simulation. Right?
If you generate independent gaussian independent samples, No is variance/Fs, where Fs is the sampling frequency.
Eb is the signal energy per bit. It would be better if you can provide some detail.
Regards

Z
 

Sorry Eb is the average bit energy, usually set as 1 in simulations.
 

Ahhh !!!!
I used both above methods i.e. No = Variance/Fs
and method told by Ahmed but alasssssss... my results are still same. (

i am stuck with it from last 2 weeks and now am feeling like dead....
 

Ghalzai said:
so Eb is required to calculate it. How to calculate Eb?? :O
...
So my number of bits sending in symbol are varying from 2-7.
Click to expand...
You can calculate Eb in this way:

Eb = (average_signal_power)*(time_per_bit) = (average_signal_power) / (bit_rate)

Equivalently, you could consider the energy per symbol an then the bits per symbol:

Es = (average_signal_power)*(time_per_symbol) = (average_signal_power) / (symbol_rate)
Eb = Es / (number_of_bits_per_symbol)

Ghalzai said:
but alasssssss... my results are still same. (
Click to expand...
Which are your wrong results?
Regards

Z
 

Status
Not open for further replies.

Similar threads

Menu
Log in
Register
Cookies are required to use this site. You must accept them to continue using the site. Learn more…