RAN550 - UE cannot see network

[khooi8913]

Dear all,

We are trying to test srsRAN with the Benetel RAN550 (RAN550-1V0.8.1_patch_2).
However, none of the UEs that we tested can detect the cell.

After looking up discussions on the forum/ GitHub issues, we have tuned the system accordingly, and there seems to be no real time issues related to the system. However, we are seeing UL HARQ failures.

2024-01-02T06:33:31.114679 [OFH     ] [I] Received packets: rx_total=1000 rx_early=0, rx_on_time=1000, rx_late=0
2024-01-02T06:33:31.530125 [UL-PHY0 ] [I] [   72.18] PRACH:  rssi=-78.7dB detected_preambles=[{idx=1 ta=72.66us power=-8.0dB snr=-8.0dB}] t=166.4us
2024-01-02T06:33:31.532515 [SCHED   ] [I] [   73.10] Slot decisions cell=0 t=3us (1 PDSCH, 0 PUSCHs): RAR: ra-rnti=0x7f rb=[0..3) tbs=9
2024-01-02T06:33:31.532540 [DL-PHY10] [I] [   73.10] PDCCH: rnti=0x007f ss_id=1 format=1_0 cce=0 al=4 t=19.2us
2024-01-02T06:33:31.532584 [DL-PHY10] [I] [   73.10] PDSCH: rnti=0x007f prb=[0, 3) symb=[2, 14) mod=QPSK rv=0 tbs=9 t=30.0us
2024-01-02T06:33:31.536008 [SCHED   ] [I] [   73.17] Slot decisions cell=0 t=1us (0 PDSCHs, 1 PUSCH): UL: ue=1024 rnti=0x4603 h_id=0 ss_id=1 rb=[3..6) rv=0 tbs=11 msg3_delay=7
2024-01-02T06:33:31.539290 [UL-PHY1 ] [I] [   73.17] PUSCH: rnti=0x4603 h_id=0 prb=[3, 6) symb=[0, 14) mod=QPSK rv=0 tbs=11 crc=KO iter=6.0 sinr=-24.7dB t=150.6us uci_t=0.0us ret_t=0.0us
2024-01-02T06:33:31.539545 [DL-PHY4 ] [I] [    74.4] PDCCH: rnti=0x4603 ss_id=1 format=0_0 cce=0 al=4 t=19.1us
2024-01-02T06:33:31.541516 [SCHED   ] [I] [    74.8] Slot decisions cell=0 t=2us (0 PDSCHs, 1 PUSCH): UL: ue=1024 rnti=0x4603 h_id=0 ss_id=1 rb=[3..6) rv=0 tbs=11 k2=4
2024-01-02T06:33:31.544773 [UL-PHY1 ] [I] [    74.8] PUSCH: rnti=0x4603 h_id=0 prb=[3, 6) symb=[0, 14) mod=QPSK rv=0 tbs=11 crc=KO iter=6.0 sinr=-23.5dB t=146.1us uci_t=0.0us ret_t=0.0us
2024-01-02T06:33:31.545038 [DL-PHY15] [I] [   74.15] PDCCH: rnti=0x4603 ss_id=1 format=0_0 cce=0 al=4 t=18.2us
2024-01-02T06:33:31.547016 [SCHED   ] [I] [   74.19] Slot decisions cell=0 t=2us (0 PDSCHs, 1 PUSCH): UL: ue=1024 rnti=0x4603 h_id=0 ss_id=1 rb=[3..6) rv=0 tbs=11 k2=4
2024-01-02T06:33:31.550276 [UL-PHY1 ] [I] [   74.19] PUSCH: rnti=0x4603 h_id=0 prb=[3, 6) symb=[0, 14) mod=QPSK rv=0 tbs=11 crc=KO iter=6.0 sinr=-20.6dB t=150.7us uci_t=0.0us ret_t=0.0us
2024-01-02T06:33:31.550535 [DL-PHY6 ] [I] [    75.6] PDCCH: rnti=0x4603 ss_id=1 format=0_0 cce=0 al=4 t=18.3us
2024-01-02T06:33:31.556020 [SCHED   ] [I] [   75.17] Slot decisions cell=0 t=1us (0 PDSCHs, 1 PUSCH): UL: ue=1024 rnti=0x4603 h_id=0 ss_id=1 rb=[3..6) rv=0 tbs=11 k2=11
2024-01-02T06:33:31.559278 [UL-PHY2 ] [I] [   75.17] PUSCH: rnti=0x4603 h_id=0 prb=[3, 6) symb=[0, 14) mod=QPSK rv=0 tbs=11 crc=KO iter=6.0 sinr=-25.7dB t=152.2us uci_t=0.0us ret_t=0.0us
2024-01-02T06:33:31.559541 [DL-PHY4 ] [I] [    76.4] PDCCH: rnti=0x4603 ss_id=1 format=0_0 cce=0 al=4 t=18.1us
2024-01-02T06:33:31.561515 [SCHED   ] [I] [    76.8] Slot decisions cell=0 t=2us (0 PDSCHs, 1 PUSCH): UL: ue=1024 rnti=0x4603 h_id=0 ss_id=1 rb=[3..6) rv=0 tbs=11 k2=4
2024-01-02T06:33:31.564769 [UL-PHY2 ] [I] [    76.8] PUSCH: rnti=0x4603 h_id=0 prb=[3, 6) symb=[0, 14) mod=QPSK rv=0 tbs=11 crc=KO iter=6.0 sinr=-19.9dB t=144.0us uci_t=0.0us ret_t=0.0us
2024-01-02T06:33:31.565016 [SCHED   ] [I] [   76.15] UL HARQ rnti=0x4603 cell=0 h_id=0: Discarding HARQ with tbs=11. Cause: Maximum number of reTxs 4 exceeded
2024-01-02T06:33:32.114684 [OFH     ] [I] Received packets: rx_total=1070 rx_early=0, rx_on_time=1070, rx_late=0

Anyone has any idea to resolve this? Tuning IQ scaling does not seem to work.
Please see the system configuration in the sections below.
Thanks!

System Information/ Configuration

Ubuntu 22.04.3 LTS (Kernel version 5.15.0-91-lowlatency)
2 x Intel(R) Xeon(R) Gold 6334 CPU @ 3.60GHz (Hyperthreading disabled)
256GB ECC DDR4 RAM
Intel E810-XXVDA4TGG1 (Attached to NUMA Node 1)

Note: RAN550 is directly connected to the server which also acts as the PTP grandmaster.

user@server:~$ lscpu -p
# The following is the parsable format, which can be fed to other
# programs. Each different item in every column has an unique ID
# starting usually from zero.
# CPU,Core,Socket,Node,,L1d,L1i,L2,L3
0,0,0,0,,0,0,0,0
1,1,0,0,,1,1,1,0
2,2,0,0,,2,2,2,0
3,3,0,0,,3,3,3,0
4,4,0,0,,4,4,4,0
5,5,0,0,,5,5,5,0
6,6,0,0,,6,6,6,0
7,7,0,0,,7,7,7,0
8,8,1,1,,32,32,32,1
9,9,1,1,,33,33,33,1
10,10,1,1,,34,34,34,1
11,11,1,1,,35,35,35,1
12,12,1,1,,36,36,36,1
13,13,1,1,,37,37,37,1
14,14,1,1,,38,38,38,1
15,15,1,1,,39,39,39,1

Configuration/ Log Files

• ran550.yml.txt
• gnb.log.txt
• RU configuration used is similar to the tutorial except the MAC addresses.
• RU radio config: 2x2 (Antenna 1 and 3), centre frequency at 3,42GHz (n78)
• TDD pattern (similar to the tutorials)

<xml>
    <tdd-config version="1.0">
            <description>DDDDDDDSUU</description>
                        <pattern index="1">
                                <slots>DDDD</slots>
                                <numerology>1</numerology>
                                <periodicity>2</periodicity>
                                <special-slot-symbols></special-slot-symbols>
                        </pattern>
                        <pattern index="2">
                                <slots>DDSUUU</slots>
                                <numerology>1</numerology>
                                <periodicity>3</periodicity>
                                <special-slot-symbols>DDDDDDGGGGUUUU</special-slot-symbols>
                        </pattern>
    </tdd-config>
</xml>

[ismagom]

Hi @khooi8913,

There seems to be an issie qith PRACH detection, see the detected time offset (ta field):

PRACH:  rssi=-78.7dB detected_preambles=[{idx=1 ta=72.66us power=-8.0dB snr=-8.0dB}]

This should be much closer to zero. What UE are you using?

Also, I think we usually test the 4x2 mode in Benetel, not sure the others work as well.

[khooi8913]

Hi @ismagom ,

We switched the RU to 4x2 mode, and we tested with several UEs:

1. Quectel RM500Q-GL
2. OnePlus 8 Pro
3. OnePlus 11
4. OnePlus Nord 3

It seems to be exhibiting a similar behavior too (I am seeing ta values from 0.76~72++).
These UEs did work with the srsRAN+USRP earlier, and also other stacks.

Any potential pointers on what else we should tune?
Thanks!

[khooi8913]

Update on our status:

1. Instead of 20MHz, we tried with larger bandwidths like 80 and 100MHZ (and nearly maxed the power for SSB block, p0 nominal for PUCCH and PUSCH). The PRACH detection issue went away, and our UE was able to see the network and attach to it. We are not quite sure what is happening here. We only tried with SISO, not MIMO. From our observation, our MediaTek-based OnePlus Nord3 UE can attach quite reliably to the network, but not the Qualcomm-based ones (OnePlus 8 Pro and OnePlus 11).

2. During transmission, the PUSCH and PDSCH error rates remain high, >50% for most of the time, and RLF happens quite often.

Are there pointers on which other parameters that we could tune to make smaller bandwidths work, and also reduce the PHY error rates?
Thanks!

[khooi8913]

Issue resolved, the UE can now also attach with smaller bandwidths.
There was a misunderstanding on the center frequency configuration on the RU, the current v0.8.1 RU firmware defaults to 100MHz (273 PRBs) and srsRAN configuration (non-100MHz), i.e., PRB numbers do not match up and are not aligned at the right place in non-100MHz bandwidths. Thus, the configured DU ARFCN needs to consider an "offset" from the RU's center frequency.

However, we are still facing very high error rates at DL/UL.
The issue will be raised/ followed up on other similar threads.

[aibtw]

I am using a similar setup with the RAN550 and E810 based NIC, and, facing similar issues.

However, it seems that my problem is in vlan tagging without a switch.
Have you found a workaround for that? That would be really helpful.

[khooi8913]

We are also not using a FH switch (i.e., E810-RAN550 direct connection), but we never faced any issues with VLAN tagging.
Maybe the driver/FW can be the issue? The driver/FW version we are using for the E810 is ice-1.12.7, and NVM 4.31.

[aibtw]

Thank you, that's really helpful!
I Just want to confirm, the only thing you did in this regard is to match vlan tag on srs configs file with vlan on RU configs file, without setting an actual vlan. Am I correct?

[khooi8913]

Yes, that is correct. No VLAN configuration on the NIC.
Only on the SRS and RU configs.

[yc541]

Hi @khooi8913, we are working on a similar setup without a FH switch but our NIC is not SyncE capable, is it still possible to sync DU/RU via PTP? Many thanks.

[khooi8913]

Hey @yc541, I believe it should work as long as your NIC can do PTP HW timestamps.
We are not using the SyncE feature on our NIC even though there is such functionality.

[yc541]

Thanks for your quick reply! Our NIC can only do SW timestamps. Guess we either need a better NIC or a FH switch.