Fwd: Limited number of HARQ process IDs

[Biljana B.]

Hi Maria, 

I am forwarding this email to the 5g-LENA users list. 

ns-developers list has a different purpose. 

For questions about 5G-LENA please use the 5G-LENA Google users list, and for more ns-3 related questions the ns-3 Google users list.

Let's continue the discussion in the 5G-LENA thread.

Biljana

---------- Forwarded message ---------
From: mariavelazq...@gmail.com <Unknown>
Date: Wednesday, September 4, 2024 at 4:36:19 PM UTC+2
Subject: Limited number of HARQ process IDs
To: ns-developers <Unknown>

  I’ve been exploring and documenting the HARQ protocol implementation in 5G-LENA using a simple gNodeB-UE simulation with the parameters listed below to make packet-level measurements. I’ve encountered a question regarding the HARQ processes: Why are only 5 HARQ process IDs used simultaneously, even though, in theory, each UE should have 16 available HARQ processes? 

*The 70 mbps rate appears to be the maximum that the configuration can handle without incurring losses.  

   double simTime = 2.5;               // seconds

   double udpAppStartTime = 0.5;       // seconds

    double rate = 70000000;     // (in bps)
    uint32_t udpPacketSize = 1460;
    double IPDT = static_cast<double>(1460 * 8) / rate;
    // other simulation parameters default values
    uint16_t numerology = 0;

    uint16_t gNbNum = 1;
    uint16_t ueNumPergNb = 1;

    double centralFrequency = 3.5e9;
    double bandwidth = 40e6;
   
    double txPower = 10;  

    uint8_t fixedMcs = 28;
    bool useFixedMcs = true;
    bool singleUeTopology = true;

[Kent Huns]

>> Why are only 5 HARQ process IDs used simultaneously, even though, in theory, each UE should have 16 available HARQ processes? 

I suppose it depends on the period(N1delay? k1delay?)  between 1st-trans and re-trans.

If it's short enough, no needs to use all IDs.