Forums

mapping symbols onto carrier waves? ifft?

Started by Johnny July 19, 2007
so i'm venturing into communications research and i'm stuck on how to
change symbols into the time domain. i have modulated bits, say [0 1]
which in bpsk is just -1 and 1. how do i take this data and put it on
the time domain to be transmitted?

it should just be a sine wave and then at one point be shifted over
180 degrees. it was my understanding that you throw the IFFT in there
to accomplish this, however, throwing in -1 and 1 into matlab gives me
2 points, which is hardly a time domain representation.

perhaps i am not understanding how the IFFT works period, is there
some way i can turn these into sine waves?

all the sample .m files do not really map symbols onto carrier waves,
and the ones that claim they do just throw it into the ifft and just
say it is in the time domain, but when moving the data down to a
couple bits and changing the modulation to phase shift keying, the
amplitude changes. and the waveform is not a simple sine wave at all.
For all my comm. research I simulate my digital comm systems in their
baseband, where you don't need to generate the corresponding sinewaves. For
e.g. in the basic AWGN channel model,

y = x + w

where y is the received time-domain (TD) signal, x is the transmitted TD
signal and w is the gaussian noise.

Perform symbol detection with y. To measure the bite-error-rate (BER) you
only need to compare the detected bits with the corresponding TX bits in x.
This method will allow you to reproduce the BER curves from any good digital
comm. text book like Proakis.

HTH
Sameer
On 7/18/07, Johnny wrote:
>
> so i'm venturing into communications research and i'm stuck on how to
> change symbols into the time domain. i have modulated bits, say [0 1]
> which in bpsk is just -1 and 1. how do i take this data and put it on
> the time domain to be transmitted?
>
> it should just be a sine wave and then at one point be shifted over
> 180 degrees. it was my understanding that you throw the IFFT in there
> to accomplish this, however, throwing in -1 and 1 into matlab gives me
> 2 points, which is hardly a time domain representation.
>
> perhaps i am not understanding how the IFFT works period, is there
> some way i can turn these into sine waves?
>
> all the sample .m files do not really map symbols onto carrier waves,
> and the ones that claim they do just throw it into the ifft and just
> say it is in the time domain, but when moving the data down to a
> couple bits and changing the modulation to phase shift keying, the
> amplitude changes. and the waveform is not a simple sine wave at all.
>
>
>