[srslte-users] pdsch_enodeb > file and synch_file example

Mon May 16 22:26:03 UTC 2016

Ismael,

Thanks, yeah I definitely missed the 3/4 sampling thing.

However, using

$ FFTLEN=384; SLOTLEN=$((15*384)); FRAMELEN=$((10*SLOTLEN)); ./synch_file -l $FRAMELEN -s $FFTLEN -t 2 -i 25prb

on the output from

./pdsch_enodeb -o 25prb -p 25 -n 100 -c 123

still produced no results.

I'm confused about the point of 3/4 sampling, wouldn't we incur a penalty by not using a power of 2 FFT? What is the benefit?

It also seems weird that synch_file works fine on a 6 PRB enodeb signal, but if I resample, say, 25 PRB down to the equivalent of 6 PRB (by filtering and decimating by 4 to go from 7.68 MHz -> 1.92 MHz), then it does not work.

Am I missing something else besides the 3/4 sampling thing?

Thanks,
-Doug

________________________________
From: Ismael Gomez [ismael.gomez at softwareradiosystems.com]
Sent: Monday, May 16, 2016 3:43 PM
To: Anderson, Douglas J.; srslte-users at lists.softwareradiosystems.com
Subject: Re: [srslte-users] pdsch_enodeb > file and synch_file example

Hi Douglas,

The problem could be that the DFT size for the 5 MHz signal is 384 instead of 512. See srslte/lib/common/phy_common.c function srslte_symbol_sz() to see the DFT sizes when 3/4 sampling is enabled (enabled by default).

So try again this synch_file program with FFTLEN=384 and SLOTLEN=15*384

cheers

On Mon, 16 May 2016 at 23:07 Anderson, Douglas J. <danderson at its.bldrdoc.gov<mailto:danderson at its.bldrdoc.gov>> wrote:
Hi,

I've been trying to figure out why I can't get any srslte library functions to work on output from pdsch_enodeb for greater than 6 PRBs.

For example here I create 100 frames at 6 PRBs, and synch_file detects all 100 frames (if I reduce the threshold). When I try the exact same thing with 25 PRBs (or any other bandwidth), I get nothing.

Peaking around with GDB, I can see that the peak to side-lobe ratio returned by the pss_synch function is ~1 for all N_id_2 above 6 PRB.

I'm concerned that it has something to do with the output of pdsch_enodeb, since I get the same results (no PSS correlation) even if I resample the output file down to 1.92 MHz before running synch_file.

Here's an example of what I'm talking about, running on srslte HEAD:

$ ./pdsch_enodeb -o 6prb -p 6 -n 100 -c 123
linux; GNU C++ version 5.3.1 20160413; Boost_105800; UHD_003.010.git-202-g9e0861e1

 - Resource Allocation Type:        Type 0
   + Resource Block Group Size:        1
   + RBG Bitmap:            0x3f
 - Modulation and coding scheme index:    1
 - HARQ process:            0
 - New data indicator:            No
 - Redundancy version:            0
 - TPC command for PUCCH:        --
 - PRB Bitmap Assignment 0st slot:
0, 1, 2, 3, 4, 5,
 - PRB Bitmap Assignment 1st slot:
0, 1, 2, 3, 4, 5,
 - Number of PRBs:            6
 - Modulation type:            QPSK
 - Transport block size:        208
Type new MCS index and press Enter: Using Volk machine: avx2_64_mmx
Done

$ FFTLEN=128; SLOTLEN=960; FRAMELEN=$((10*SLOTLEN)); ./synch_file -l $FRAMELEN -s $FFTLEN -t 5 -i 6prb
linux; GNU C++ version 5.3.1 20160413; Boost_105800; UHD_003.010.git-202-g9e0861e1
Initializing...Using Volk machine: avx2_64_mmx
done in 0 s 638 ms

    Fr.Cnt    N_id_2    N_id_1    Subf    PSS Peak/Avg    Idx    m0    m1    CFO
    ===============================================================================
    1    0    41    5    6.413        960    13    11    -0.000
    2    0    41    0    7.155        960    11    13    -0.000
[...]
    98    0    41    0    9.678        960    11    13    -0.000
    99    0    41    5    8.089        960    13    11    -0.000

Average exec time: 1.553 ms / frame. 6.182 Msamp/s (31.056% CPU)
Average CFO: -0.000
Done

$ ./pdsch_enodeb -o 25prb -p 25 -n 100 -c 123
linux; GNU C++ version 5.3.1 20160413; Boost_105800; UHD_003.010.git-202-g9e0861e1

 - Resource Allocation Type:        Type 0
   + Resource Block Group Size:        2
   + RBG Bitmap:            0x1fff
 - Modulation and coding scheme index:    1
 - HARQ process:            0
 - New data indicator:            No
 - Redundancy version:            0
 - TPC command for PUCCH:        --
 - PRB Bitmap Assignment 0st slot:
0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24,
 - PRB Bitmap Assignment 1st slot:
0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24,
 - Number of PRBs:            25
 - Modulation type:            QPSK
 - Transport block size:        904
Type new MCS index and press Enter: Using Volk machine: avx2_64_mmx
Done

$ FFTLEN=512; SLOTLEN=3840; FRAMELEN=$((10*SLOTLEN)); ./synch_file -l $FRAMELEN -s $FFTLEN -t 5 -i 25prb
linux; GNU C++ version 5.3.1 20160413; Boost_105800; UHD_003.010.git-202-g9e0861e1

Initializing...Using Volk machine: avx2_64_mmx
done in 2 s 709 ms

    Fr.Cnt    N_id_2    N_id_1    Subf    PSS Peak/Avg    Idx    m0    m1    CFO
    ===============================================================================

Average exec time: 9.369 ms / frame. 4.098 Msamp/s (46.847% CPU)
Average CFO: 0.000
Done
_______________________________________________
srslte-users mailing list
srslte-users at lists.softwareradiosystems.com<mailto:srslte-users at lists.softwareradiosystems.com>
http://www.softwareradiosystems.com/mailman/listinfo/srslte-users
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.srsran.com/pipermail/srsran-users/attachments/20160516/3e8da114/attachment.htm>