Introducing NVIDIA Aerial Research Cloud for Innovations in 5G and 6G

Originally published at: https://developer.nvidia.com/blog/introducing-aerial-research-cloud-for-innovations-in-5g-and-6g/

At NVIDIA GTC 2023, NVIDIA introduced Aerial Research Cloud, the first fully programmable 5G and 6G network research sandbox, which enables researchers to rapidly simulate, prototype, and benchmark innovative new software deployed through over-the-air networks. The platform democratizes 6G innovations with a full-stack, C-programmable 5G network, and jumpstarts ML in advanced wireless communications using NVIDIA-accelerated…

1 Like

Hi!
do you have any limitations because of latencies issues in your system?
I mean, 5G 1ms round trip time seems to be too tight to reach for a GPU architecture

Do you support 120kHz subcarrier spacing? Thank you!

1 Like

120kHz currently not supported E2E

@fredrik.molander - can we use this towards developer forum

ARC 1.2 is available and tested for Dell R750 + A100X. Testing is not complete for Gigabyte + A100 and expected to close end of the month

2 Likes

Yes lets use the forum.

2 Likes

If there are previously setup 1-1 Slack channels, expect these to become inactive starting March 15th 2024. Please transition to using the developer forum for Q/A

What firmware versions of the foxconn O-RU have been tested with aerial v1.1?

  • I updated the O-RU to v2.6.9 eng → 2.6.9 oam. For some reason ssh isn’t shown in the process list. This is unexpected.

*I am also getting the following clgc_status, and m-plane fault status messages that seem to be new.

ptp: rms=2 max=4 freq=-348 delay=84
10R: sec=7 hps=7 64b=6172 65to128=6027 total=12258 uni=46 uni>1158=1 multi=12187 crc_err=0
10T: sec=7 hps=7 64b=2596 65to128=76 total=2719 uni=54 uni>1158=0 multi=2638 crc_err=0 state=1 start=0 adj=-2 rstcnt=0
xRN: total=0 c_early=0 c_on=0 c_late=0 err1=0 err2=0 err3=0 err4=0 drop=0
Latch later 1pps time=46934443 swi4010=46934443 xran_sec=46934441 acc_diff[8]=-2 hps_sec=7 cur_sec=7 tx_through=1 rx_through=3
Tx Att = 13000 13700 13700 13700
TX1 clgc_status 0x340c
TX2 clgc_status 0x340c
TX3 clgc_status 0x340c
TX4 clgc_status 0x340c
Trigger M-plane fault_id 28 info (16, 2, 0)
Hold M-plane fault_id 28 info (16, 2, 0) flag is occupied

*The temperature is rising much faster than expected.

Trying to run the latest version 3.1.15 eng → oam

  • init_rrh_config_enable_cuplane doesn’t give the same log messages as ver 2.6.9 so if anyone has experience with using it that would be helpful.

We will transition the foxconn radios to version 3.1.15 and progress forward. Since the serial messages are different with 3.1.15 we will find the location of the log to tail.

1 Like

One year after introducing ARC, this year please register/forward/join us for a GTC panel discussion on 5G and 6G programmable cellular networks. Link below. Programmable 5G and 6G cellular network

1 Like

HI Andrew,
We have tested with both 2.6.9 and 3.1.15.xxx
With fw 3.1.15. the log is printed to /var/log/rrh_log_print/rrh_log_print.log.
Regarding the ssh and errors i think we need to involve Foxconn.
Regards
Fredrik

Thanks, when upgrading to version 3.1.15 ssh service seems to be ok. I found the logs, and the ~/test/ch_con.sh and ~/test/ptp_con.sh seem to be helpful. I’ll send Foxconn my logs, but do you happen to see the dpdErrorCode 340c or TX4 can not calculate tx power, dpdErrorCode = 0x340c? It is probably because I am not running L1 yet. I think 340c error code means that the “PA did not transmit anything”

[2024-02-29 18:16:01.040] 10R: sec=384 hps=384 64b=12389 65to128=18797 total=31244 uni=1811 uni_gt_1158=0 multi=29421 crc_err=0
[2024-02-29 18:16:01.040] 10T: sec=384 hps=384 64b=9985 65to128=487 total=11851 uni=1801 uni_gt_1158=565 multi=10004 crc_err=0 state=1 start=0 atick=0 iatick=768001 adj=-34 c_rst=0 x_rst=0
[2024-02-29 18:16:01.040] xRN: total=0 c_early=0 c_on=0 c_late=0 err_tci=0 err_ecpri=0 err_port=0 err_sct=0 err_total=0
[2024-02-29 18:16:01.040] RRH_state=READY_FOR_DATA
[2024-02-29 18:16:01.040] ptp: state=3 sec=384 rms=4 max=8 freq=-122 delay=187
[2024-02-29 18:16:01.091] proc info 0_min:2166 1_min:8146 , 0_avg:2292 1_avg:8347 , 0_max:3109 1_max:9918
[2024-02-29 18:16:01.091] Latch later 1pps=0xf21aa03 swi4010=0xf21aa03 xran_sec=0xf21aa03 1pps_d=122880000 swi4010_1pps_d=0 swi4010_d=122880000 swi4010_xran_d=0 acc_diff[5]=-34 hps_sec=384 cur_sec=384 tx_through=4 rx_through=3
[2024-02-29 18:16:01.141] dpdErrorCode 0x340c 0x340c 0x340c 0x340c
[2024-02-29 18:16:01.141] dpdDirectEvm 0 0 0 0 m%
[2024-02-29 18:16:01.141] dpdIndirectEvm 0 0 0 0 m%
[2024-02-29 18:16:01.141] dpdMeanOrxPower -2147483648 -2147483648 -2147483648 -2147483648 mdBFS
[2024-02-29 18:16:01.141] dpdPeakOrxPower -2147483648 -2147483648 -2147483648 -2147483648 mdBFS
[2024-02-29 18:16:01.141] dpdMeanTuPower -2147483648 -2147483648 -2147483648 -2147483648 mdBFS
[2024-02-29 18:16:01.141] dpdPeakTuPower -2147483648 -2147483648 -2147483648 -2147483648 mdBFS
[2024-02-29 18:16:01.141] dpdMeanTxPower -2147483648 -2147483648 -2147483648 -2147483648 mdBFS
[2024-02-29 18:16:01.141] dpdPeakTxPower -2147483648 -2147483648 -2147483648 -2147483648 mdBFS
[2024-02-29 18:16:01.141] TX1 clgcTxRmsPower=0
[2024-02-29 18:16:01.141] TX1 clgcOrxRmsPower=64
[2024-02-29 18:16:01.141] TX2 clgcTxRmsPower=0
[2024-02-29 18:16:01.141] TX2 clgcOrxRmsPower=62
[2024-02-29 18:16:01.141] TX3 clgcTxRmsPower=0
[2024-02-29 18:16:01.141] TX3 clgcOrxRmsPower=60
[2024-02-29 18:16:01.141] TX4 clgcTxRmsPower=0
[2024-02-29 18:16:01.141] TX4 clgcOrxRmsPower=62
[2024-02-29 18:16:01.141] TX1 can not calculate tx power, dpdErrorCode = 0x340c
[2024-02-29 18:16:01.141] TX2 can not calculate tx power, dpdErrorCode = 0x340c
[2024-02-29 18:16:01.141] TX3 can not calculate tx power, dpdErrorCode = 0x340c
[2024-02-29 18:16:01.191] TX4 can not calculate tx power, dpdErrorCode = 0x340c
[2024-02-29 18:16:01.192] xRN: u-plane u_early=0 u_on=0 u_late=0
[2024-02-29 18:16:01.192] Temperature of ad9025 is 66 degree Celsius
[2024-02-29 18:16:01.192] curr mdBFS of Rx = 56000 55500 55500 56000
[2024-02-29 18:16:01.192] keep REG_AD9025_RX_BYPASS_ADDR to 0x00000f, small_power_cnt = 0
[2024-02-29 18:16:01.192] fpga REG_XRAN_SYM_DL_IQ_SCALING : 0x000003

Hi There, I would like to know the whether aerial sdk will support for L40S GPU since A100 has discontinued from January 2024.

Hi all,

is there some documentation and guidelines available with the steps to rebuild NVIDIA Aerial L1 and docker image for oai-gnb-aerial after making changes in the source code?

Also, is ARC 1.2 compatible with the newest version of OAI core network?

1 Like

We are testing ARC with the Foxconn RU RPQN-4800E in the CBRS band. We have faced an issue that when ARC runs, the logs on the RU do not show any data activity (no control or data plane packets are received or transmitted) and none of the TX antennas are on (dpdErrorCode = 0x340c on all 4 TX antennas). How can we monitor the front-haul traffic when the ARC is on?

HI
For building the NVIDIA Aerial L1 source code please follow the directions here
Part 6. Validate the Setup - NVIDIA Docs.

ARC 1.2 supports Aerial L1 23-4 and OAI_Aerial_v2.2.2.

1 Like

Thank you very much Fredrick! We will try these directions.