How to disable HARQ

[Guy Gabay]

Dear LENA 5G team,

I’m currently working with the LENA 5G simulation framework and would like to ask whether there is a way to disable HARQ (Hybrid Automatic Repeat reQuest) in the simulations.

Is there a specific configuration parameter or class I should modify to turn it off?

Any guidance or pointers would be greatly appreciated.

Best regards

[Kent Huns]

Check some examples.

    nrHelper->SetSchedulerAttribute("EnableHarqReTx", BooleanValue(false));

--
You received this message because you are subscribed to the Google Groups "5G-LENA-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to 5g-lena-user...@googlegroups.com.
To view this discussion, visit https://groups.google.com/d/msgid/5g-lena-users/c7a36016-c99c-4606-b6f3-fe4c9cd027abn%40googlegroups.com.

[Guy Gabay]

Thank you for the guidance regarding the use of:

nrHelper->SetSchedulerAttribute("EnableHarqReTx", BooleanValue(false));

I tried setting this attribute as suggested in order to disable HARQ retransmissions. However, while tracing the following source:

Config::Connect("/NodeList/*/DeviceList/*/$ns3::NrNetDevice/$ns3::NrGnbNetDevice/BandwidthPartMap/*/NrGnbMac/UlScheduling", MakeCallback(&ReportUlPhyResourceBlocksEnb0));

I noticed that the parm.m_harqId value is still being set, mostly to 14 or 15, rather than 0. This seems to indicate that HARQ processes are still active, or at least being assigned.

Is this expected behavior even when EnableHarqReTx is set to false? Or is there an additional configuration or class I need to modify to fully disable HARQ operation (including allocation of HARQ process IDs)?

Any clarification would be greatly appreciated.

 

ב-יום שישי, 6 ביוני 2025 בשעה 06:09:42 UTC-4, Kent Huns כתב/ה:

[Biljana B.]

Hi Guy Gabay, 

sorry for the delay. I am sorry, but in 5G-LENA module HARQ cannot be completely disabled. In the current code it forms part of the core scheduler's code, and at the time it was designed it was not prepared to be possible to disable it completely. We were also interested in some simulation campaigns to disable it, just in order to do comparisons, but it was not trivial, and we did not have time to implement such an option, even if it would be really useful for research studies. 

The parameter that Kent Huns shared with you is to disable retransmissions, but still HARQ processes are going to exist and consequently some delays due to timeouts will be experienced, so it is also not trivial to interpret such mode.

If you will develop a modular/configurable and easy to integrate option for enabling/disabling HARQ, it could be interesting and valuable contribution to 5G-LENA.

Kind regards,

Biljana

[Guy Gabay]

From what I understand, the closest way to achieve a “no HARQ” behavior without modifying the core codes is by configuring the following:

• Setting EnableHarqReTx = false to disable retransmissions,

• Limiting NumHarqProcess = 1 to simplify the HARQ process and avoid parallel handling,

• Using a custom TDD pattern to minimize the delay between UL and DL, reducing the waiting time for feedback.

Based on my observations, this setup seems to approximate the behavior of a system without HARQ — both in terms of throughput and latency.

Is this approach considered valid and commonly used in the community to simulate a no-HARQ scenario?
Are there any important caveats or alternative recommendations I should be aware of?

I’d really appreciate your insights.

ב-יום שלישי, 17 ביוני 2025 בשעה 12:51:11 UTC-4, Biljana B. כתב/ה:

[Kent Huns]

>> The parameter that Kent Huns shared with you is to disable retransmissions, but still HARQ processes are going to exist and consequently some delays due to timeouts will be experienced

Thank you for the explanation, Biljana. I've never found any difficulty in evaluating delay times, but didn't know that.

>> Limiting NumHarqProcess = 1 to simplify the HARQ process and avoid parallel handling

Perhaps the correct answer is the opposite. If enough parallel processes are set, the N1 delays can canceled and it enables to continue the 1st transmission without any delay.

https://groups.google.com/g/5g-lena-users/c/x-odUmXS_vE/m/dEi9s2ivAQAJ