[LBT / Wifi Coexistance module] : 802.11ac Wifi throughput is higer than defined by standard

[saumil shah]

Dear all,

I am working on coexistence study of LTE / Wifi in unlicensed band. Currently i am working on "laa-wifi-simple" example. Now i am interested to working with IEEE 802.11ac standard for wifi , i was making some test to make sure it works correct with coexistence module as default standard is IEEE 802.11n in 5GHz.

I have observed that when i am using 802.11ac standard i am getting higher throughput than defined by the standard.

I have made below changes in the scenario-helper.cc file.

1) Changed the standard from 802.11n_5GHZ to 802.11ac in ConfigureWifiSta() and ConfigureWifiAp().

wifi.SetStandard (WIFI_PHY_STANDARD_80211ac);

2) As i am interested in SISO i have chnaged the number of transmitters and receivers in ConfigureWifiSta() and ConfigureWifiAp() to 1.(To make it SISO)

spectrumPhy.Set ("Receivers", UintegerValue (1));
spectrumPhy.Set ("Transmitters", UintegerValue (1));

I am running the script as below.
./waf --run "src/laa-wifi-coexistence/examples/laa-wifi-simple --duration=10 --d2=1000 --RngRun=3 --RngSeed=7 --cellConfigA=Wifi  --udpPacketSize=1472 --pcapEnabled=true --udpRate=100000000"

With above script for standard 802.11ac i am getting below output.
--------monitorA----------
Flow 1 (11.0.0.1:49153 -> 17.0.0.2:9) proto UDP
  Tx Packets: 84457
  Tx Bytes:   126685500
  TxOffered:  101.348 Mbps
  Rx Bytes:   126681000
  Throughput: 101.902 Mbps
  Mean delay:  0.347303 ms
  Mean jitter:  0.124728 ms
  Duration:     9.94536 s
  Rx Packets: 84454
--------monitorB----------
Flow 1 (12.0.0.1:49153 -> 18.0.0.2:9) proto UDP
  Tx Packets: 84264
  Tx Bytes:   126396000
  TxOffered:  101.117 Mbps
  Rx Bytes:   126393000
  Throughput: 101.901 Mbps
  Mean delay:  0.346851 ms
  Mean jitter:  0.124572 ms
  Duration:     9.92285 s
  Rx Packets: 84262

As we can see throughput is 101 Mb/s but for SISO mode with 20 MHz bandwidth max physical data rate achieved by 802.11ac standard is 78 Mb/s.
After one point whatever high application data rate we will give throughput must saturate but here it is not saturating and as you increase application data rate throughput will also increase.

But with default standard 802.11n 5GHZ i am getting correct throughput and after one point irrespective of application data rate throughput will saturate which is around 60 Mb/s which is correct. Below is the output with standard 802.11n in 5GHz.

--------monitorA----------
Flow 1 (11.0.0.1:49153 -> 17.0.0.2:9) proto UDP
  Tx Packets: 84457
  Tx Bytes:   126685500
  TxOffered:  101.348 Mbps
  Rx Bytes:   75363000
  Throughput: 60.621 Mbps
  Mean delay:  80.0319 ms
  Mean jitter:  0.15538 ms
  Duration:     9.94546 s
  Rx Packets: 50242
--------monitorB----------
Flow 1 (12.0.0.1:49153 -> 18.0.0.2:9) proto UDP
  Tx Packets: 84264
  Tx Bytes:   126396000
  TxOffered:  101.117 Mbps
  Rx Bytes:   75190500
  Throughput: 60.6199 Mbps
  Mean delay:  80.0378 ms
  Mean jitter:  0.155382 ms
  Duration:     9.92288 s
  Rx Packets: 50127

Questions:

1) Does LBT / Wifi coexistance module not support any other standard than 802.11n 5 Ghz ?

2) If coexistence module support other standards then what can be the reason behind above mentioned behavior for 802.11ac standard?

3) Should i need to modify something else to change the standard?

Best Regards
Saumil

[Tom Henderson]

Thanks for a careful email report that made it easy for me to reproduce your simulations.

LBT/Wi-Fi has only been tested with 11n.  802.11ac should be compatible.  However, as you noted, you have to edit the helper code to enable it, and there is a statement that is not providing the intended effect.

In the helper code, you changed the first statement to use 11ac, and you later can see a statement that sets the channel width to 20 MHz in the SpectrumWifiPhy helper:

  wifi.SetStandard (WIFI_PHY_STANDARD_80211ac);
  WifiMacHelper mac;

  spectrumPhy.Set ("ShortGuardEnabled", BooleanValue (false));
  spectrumPhy.Set ("ChannelWidth", UintegerValue (20));
  mac.SetType ("ns3::StaWifiMac",
               "Ssid", SsidValue (ssid),
               "ActiveProbing", BooleanValue (false));

  for (uint32_t i = 0; i < ueNodes.GetN (); i++)
    {
      //uint32_t channelNumber = 36 + 4 * (i%4);
      uint32_t channelNumber = 36;
      spectrumPhy.SetChannelNumber (channelNumber);
      staDevices.Add (wifi.Install (spectrumPhy, mac, ueNodes.Get (i)));
    }


However, there is a subtlety that is preventing that 20MHz configuration from working as you might expect.  SetStandard (80211ac) overrides the 20 MHz setting in the Phy helper and configures an 80 MHz channel by default, when Install() is called.

This problem doesn't arise in 11n because the default channel width is already 20MHz.

To fix this, you probably have to change the configuration to better support 11ac.  For example, you can obtain a SpectrumWifiPhy pointer from the staDevices container *after* you perform the install; e.g.

  Ptr<WifiNetDevice> nd = staDevices.Get(0)->GetObject<WifiNetDevice> ();
  Ptr<SpectrumWifiPhy> swp = nd->GetPhy ()->GetObject<SpectrumWifiPhy> ();
  swp->SetChannelWidth (20);

and then you can directly change the channel on that object (and repeat for each STA and AP in the simulation).

However, since you seem to be using 20 MHz channels, I don't see much utility in trying to configure 11ac with this codebase because the advanced 11ac features like MU-MIMO are not available, and it basically amounts to having 11n with an additional MCS (MCS8).   If you don't plan to use this MCS, then I would recommend to stay with 11n configuration.

- Tom

[saumil shah]

Dear Tom,

Thank you very much for you reply and solution to solve the problem.

I have tried it and it is working correctly now for 802.11ac standard too. But yeah as you suggested i will use 802.11n only as i don't have any specific requirements for MCS8.

Best Regards
Saumil