FPGA Central

CDMA uses psuedo random sequences to seperate users sharing the same
frequency band. The standards (which I don't have access to) specify
a polynomial used to generate these sequences. The polynomial
describes where the taps go, in a linear feedback shift register.

The LFSR for a particular polynomial can be implemented in two ways,
either in a Galois form (little adders at the taps) or Fibonacci form
(a huge adder feeding the shift register). Even though they use the
same polynomial, they will produce a different bit sequence.

Do the CDMA standards specify which form to use?

If not, is it possible to use either form interchangably in a CDMA
system and still be able to isolate a specific user? i.e. does a
sequence generated by a Galois form cross-correlate well with one
generated by Fibonacci form, as long as they have the same polynomial?
I'm guessing the answer is no.

Thanks in advance.

On 14 Nov 2004 16:02:16 -0800, [email protected] (EM) wrote:

>CDMA uses psuedo random sequences to seperate users sharing the same
>frequency band. The standards (which I don't have access to) specify
>a polynomial used to generate these sequences. The polynomial
>describes where the taps go, in a linear feedback shift register.
>
>The LFSR for a particular polynomial can be implemented in two ways,
>either in a Galois form (little adders at the taps) or Fibonacci form
>(a huge adder feeding the shift register). Even though they use the
>same polynomial, they will produce a different bit sequence.
>
>Do the CDMA standards specify which form to use?
>
>If not, is it possible to use either form interchangably in a CDMA
>system and still be able to isolate a specific user? i.e. does a
>sequence generated by a Galois form cross-correlate well with one
>generated by Fibonacci form, as long as they have the same polynomial?
> I'm guessing the answer is no.

The sequences are the same! However, the internal states aren't the
same, which is probably what confused you. (Take particular care if
the CDMA standard refers to the internal state, e.g. during
initialisation.)

This thread will help:
http://groups.google.com/groups?thre...ing.google.com

Regards,
Allan

> The sequences are the same!

Ahh. I must have an error in my simulations, which produced different
sequences (that weren't even phase shifts of each other).

> This thread will help:
> http://groups.google.com/groups?thre...ing.google.com

Thanks a bunch for the link, Allan. I should have Googled outside of
comp.dsp. Silly me.