[srslte-users] Questions on non-standard sampling rates

Tom Tsou tom at tsou.cc
Wed Mar 8 23:20:10 UTC 2017


On Wed, Mar 8, 2017 at 12:18 PM, Felipe Augusto Pereira de Figueiredo
<zz4fap at gmail.com> wrote:
> I'm studying the srsLTE code and I have two questions. I noticed that the
> sampling rate for a BW of 5 MHz is 15000 * 384 = 5.76 MHz and the master
> clock is 4 * 5.76 MHz = 23.04 MHz. Checking the code I saw it is a non
> standard rate.
>
> 1) Why do you use those non-standdard sampling frequencies?
> 2) Is it possible to receive real world LTE signals by using those non
> standard sampling rates? If so, how is it possible?

Strictly speaking, sample rate and FFT size are not specified by the
standard specification. Parameters that are specified are subcarrier
spacing, subcarrier count, and symbol rate. Practicality dictates
'reasonable' FFT sizes based on radix-2 and, in certain cases, radix-3
implementations as we can use with 20, 10 and 5 MHz bandwidth
configurations.

Note that a radix-3 FFT can also be used at 3 and 1.4 MHz settings,
but the cyclic prefix length becomes a fractional, rather than whole,
number of symbols. Again, practically means that radix-2 FFT sizes are
used in these cases for simplicity of implementation.

   -TT



More information about the srsran-users mailing list