No info find in HARQ buffer for UE 0

[hou yahiya]

Hi Tommaso,

I hope you are doing well.

I'm trying to simulate an LTE scenario containing 100 UEs and 1 eNdB, but it seems impossible to create more than 35 UEs. it gives this error after compilation:

****************************************************************************************************

msg="No info find in HARQ buffer for UE 0", +0.047000000s 0 file=/home/houcem/ns-3-dev/src/lte/model/pf-ff-mac-scheduler.cc, line=630
NS_FATAL, terminating
terminate called without an active exception
Command 'build/scratch/ns3.38-LteExample-default' died with <Signals.SIGABRT: 6>.

*****************************************************************************************************

Can you suggest to me what I should do , Sir?

Best regards,

[Tommaso Pecorella]

Hi,

if you believe it's a bug please open an issue in our tracker: https://gitlab.com/nsnam/ns-3-dev/-/issues/

[Shyam Parekh]

I'm getting the same error. I'm also using ns-3.38. I see that Hou Yahiya has opened this as an issue on GitLab.

I wonder if this is related to the Idle Mode initial cell selection as discussed for the thread with the title msg= "No info find in HARQ buffer for UE". Any update on this issue? Is this fixed in ns-3.39 by any chance?

Thanks,

Shyam

[Gabriel Ferreira]

Did you guys actually configure the SRS timing?

From lena-dual-stripe.cc: "SRS Periodicity (has to be at least greater than the number of UEs per eNB)"
Valid values are {0, 2, 5, 10, 20, 40, 80, 160, 320}, as specified in 3GPP TS 36.213 table 8.2-1 UE Specific SRS Periodicity

You can use the following if you want 100 UEs in an eNB.

Config::SetDefault("ns3::LteEnbRrc::SrsPeriodicity", UintegerValue(160));

[Shyam Parekh]

Thanks Gabriel!

Tried a run with SRS Periodicity value of 320 with 200 UEs. It ran successfully!

I'm surprised as I'm currently using PUSCH based CQIs (as opposed to SRS based). When I was using SRS based CQIs, I was careful about setting SRS Periodicity appropriately, but thought that it wouldn't matter for PUSCH based CQIs.

Thanks,

Shyam

[minjeong]

I've been thinking about this for a long time, but thanks to you, I solved it.  Thanks Gabriel !!!

2023년 7월 13일 목요일 오전 4시 15분 38초 UTC+9에 Gabriel Ferreira님이 작성: