Samples _per_symbol parameter

Hi everyone,
i’am trying to do simply QAM modulation for that the source i
used
is vector source which gives 1byte of 1’s and 0’s which is fed to QAM
modulator .

Can anyone tell me how do i calculate samples/symbol .I know the
traditional formula

samples/symbol=(samples/second)/(symbols/second).

symbols/sec= (Bits/sec)/(Bits/symbol)

I having the problem to decide bits/symbol and bits/sec.Please help me

Thanks
Sandhya

Hi,
Bits/symbol depends on modulation scheme that you use to transmit.
For
BPSK, it is 1 bit/symbol, for QPSK it is 2 bit/symbol and for 64-QAM it
is
6 bits/symbol. So you can set it according to modulation scheme you use.

Samples/symbol is normally default set to 2. I am not aware in gnuradio
but
in certain other implementations samples/symbol is kept greater than 2,
say
4 or 8 to account for pulse shaping.

Hi karan ,
Thanks for reply .
well ya in gnuradio also if i set the value of samples/symbol below 2 it
is
throwing me an error .
So is there any specific reason behind using the value greater than 2?
and is that higher the value of samples/second the smoother will be the
waveform? please correct me if i’am wrong .

Thanks
Sandhya

Hi Sandhya,
You need atleast 2 samples/symbol for demodulation.
Normally when there are sharp transitions in modulated carrier, it
results
in high frequency components in the frequency domain. This is a problem
because there will be power across frequencies beyond the bandwidth of
the
signal. Hence pulse shaping is employed where the time domain waveform
is
smoothened so that all the power in the freq domain is concentrated only
within the bandwidth of signal and that prevents ISI. Hence because the
waveform transitions are smoothened, there will be a need for higher
samples/symbol to demodulate the symbol.

That was really a very clear explanation
Thank you so much.
Now i got the difference between the terms and also concept behind it

Thanks
sandhya