AP Testing with LANforge WiFIRE

This describes some of the test results when using LANforge WiFIRE to test various APs and configurations. These tests are not exhaustive and are subject to random interference from other WiFi equipment, so direct comparisons may not be so useful.

With the release of LANforge 5.2.9, the LANforge-GUI supports customized plugins and ships with the 'WiFi Capacity Test' plugin. This plugin makes it easy to do capacity testing and generate graphical reports. For more information, see the LANforge-GUI Users Guide plugins section. An example report is shown here.

Below are some test results created manually using older releases of LANforge.

1. Asus RT-N66U: Firmware 3.0.0.3.90
   This is a dual-radio 802.11a/b/g/n AP that supports up to 64 stations. It appears to have a bug that causes stations to get stuck in associated-but-unauthorized state if the EAPOL authentication packet 3-of-4 is re-sent. LANforge version 5.2.4 introduces a work-around for this as long as the stations are using DHCP instead of fixed IP addresses.

   Highlights: At least 237Mbps of UDP throughput, and supports 62 stations with good shared throughput.

   Tested with LANforge-WIFIRE CT521-400.
   Using WPA authentication unless otherwise noted.
   Using single radio on the CT521.
   AP is set up in routed mode (NAT, DHCP, etc)
   
   Test 1:  Station-to-Station: 62 stations, 256kbps UDP stream, 1472 byte payload.
        Channel 11, 2.4Ghz, HT-20
      Median one-way latency:  30ms
      Average packet-loss: 0.13%
      Total rx-rate: 16Mbps
      Median Jitter: 12ms
      NOTES:  Stable results.
   
   Test 2:  Station-to-Station:  62 stations, 384kbps UDP stream, 1472 byte payload.
        Channel 11, 2.4Ghz, HT-20
      Median one-way latency:  65ms
      Average packet-loss: 0.10%
      Total rx-rate: 24Mbps
      Mdian Jitter: 33ms
      NOTES:  Stable results, a few stations are being starved from time to time.
   
   Test 3:  Wired-to-Station:  62 stations, 384kbps UDP stream, 1472 byte payload.
        Channel 11, 2.4Ghz, HT-20
      Median one-way latency:  8ms
      Average packet-loss: 0.02%
      Total rx-rate: 24Mbps
      Median Jitter: 2ms
      NOTES:  Stable results.
   
   Test 4:  Wired-to-Station:  62 stations, 1Mbps UDP stream, 1472 byte payload.
        Channel 11, 2.4Ghz, HT-20
      Median one-way latency:  102ms
      Average packet-loss: 7%
      Total rx-rate: 58Mbps
      Median Jitter: 18ms
      NOTES:  Stable results, but too much packet loss
   
   Test 5:  Wired-to-Station:  62 stations, 1Mbps UDP stream, 1472 byte payload.
        Channel 157, 5Ghz, HT-40
      Median one-way latency:  89ms
      Average packet-loss: 0.18%
      Total rx-rate: 62Mbps
      Median Jitter: 17ms
      NOTES:  Stable results.
   
   Test 6:  Wired-to-Station:  62 stations, 1.1Mbps UDP stream, 1472 byte payload.
        Channel 157, 5Ghz, HT-40
      Median one-way latency:  182ms
      Average packet-loss: 3.65%
      Total rx-rate: 66Mbps
      Median Jitter: 17ms
      NOTES:  Stable results, but too much packet loss.
   
   Test 7:  Wired-to-Station:  10 stations, 10Mbps UDP stream, 1472 byte payload.
        Channel 157, 5Ghz, HT-40
      Median one-way latency:  8ms
      Average packet-loss: 0.17%
      Total rx-rate: 100Mbps
      Median Jitter: 1ms
      NOTES:  Stable results
   
   Test 8:  Wired-to-Station:  10 stations, 12.5Mbps UDP stream, 1472 byte payload.
        Channel 157, 5Ghz, HT-40
      Median one-way latency:  50ms
      Average packet-loss: 0.01%
      Total rx-rate: 125Mbps
      Median Jitter: 1ms
      NOTES:  Mostly stable results, but system is at maximum load.  Latency sometimes
              jumps above 500ms....
   
   Test 9:  Wired-to-Station:  10 stations, 20Mbps UDP stream, 1472 byte payload
        Channel 157, 5Ghz, HT-40, NO ENCRYPTION
      Median one-way latency:  6ms
      Average packet-loss: 0.41%
      Total rx-rate: 199Mbps
      Median Jitter: 0ms
      NOTES:  Stable results.
   
   Test 10:  Wired-to-Station:  10 stations, 25Mbps UDP stream, 1472 byte payload
        Channel 157, 5Ghz, HT-40, NO ENCRYPTION
      Median one-way latency:  24ms
      Average packet-loss: 4.84%
      Total rx-rate: 237Mbps
      Median Jitter: 0ms
      NOTES:  Mostly stable results, but too much packet loss.  This is limit of
              what this system can do.
   

2. Asus RT-N56U: Firmware 1.0.1.4o
   This is a dual-radio 802.11a/b/g/n AP that supports up to 31 stations.

   Highlights: At least 160Mbps of UDP throughput, and supports 31 stations with decent shared throughput.

   Tested with LANforge-WIFIRE CT521-400.
   Using WPA authentication unless otherwise noted.
   Using single radio on the CT521.
   AP is set up in routed mode (NAT, DHCP, etc)
   
   Test 1:  Station-to-Station: 30 stations, 256kbps UDP stream, 1472 byte payload.
        Channel 11, 2.4Ghz, HT-20
      Median one-way latency:  94ms
      Average packet-loss: 2.6%
      Total rx-rate: 7.3Mbps
      Median Jitter: 54ms
      NOTES:  Stable results, but too much packet loss.
   
   Test 2:  Station-to-Station: 30 stations, 350kbps UDP stream, 1472 byte payload.
        Channel 11, 2.4Ghz, HT-20
      Median one-way latency:  138ms
      Average packet-loss: 5.75%
      Total rx-rate: 9.6Mbps
      Median Jitter: 48ms
      NOTES:  Stable results, but too much packet loss.
   
   Test 3:  Wired-to-Station:  31 stations, 384kbps UDP stream, 1472 byte payload.
        Channel 11, 2.4Ghz, HT-20
      Median one-way latency:  8ms
      Average packet-loss: 0%
      Total rx-rate: 11.9Mbps
      Median Jitter: 1ms
      NOTES:  Stable results.
   
   Test 4:  Wired-to-Station:  31 stations, 850kbps UDP stream, 1472 byte payload.
        Channel 11, 2.4Ghz, HT-20
      Median one-way latency:  105ms
      Average packet-loss: 0%
      Total rx-rate: 26.5Mbps
      Median Jitter: 1ms
      NOTES:  Stable results, this is about the limit for this scenario.
   
   Test 5:  Wired-to-Station:  31 stations, 850kbps UDP stream, 1472 byte payload.
        Channel 157, 5Ghz, HT-40
      Median one-way latency:  2ms
      Average packet-loss: 0.02%
      Total rx-rate: 26Mbps
      Median Jitter: 0ms
      NOTES:  Stable results.
   
   Test 6:  Wired-to-Station:  31 stations, 3.2Mbps UDP stream, 1472 byte payload.
        Channel 157, 5Ghz, HT-40
      Median one-way latency:  5ms
      Average packet-loss: 0.16%
      Total rx-rate: 99Mbps
      Median Jitter: 3ms
      NOTES:  Stable results, this is about the limit for this scenario.
   
   Test 7:  Wired-to-Station:  10 stations, 10Mbps UDP stream, 1472 byte payload.
        Channel 157, 5Ghz, HT-40
      Median one-way latency:  2ms
      Average packet-loss: 0.10%
      Total rx-rate: 100Mbps
      Median Jitter: 0ms
      NOTES:  Stable results
   
   Test 8:  Wired-to-Station:  10 stations, 16Mbps UDP stream, 1472 byte payload.
        Channel 157, 5Ghz, HT-40
      Median one-way latency:  4ms
      Average packet-loss: 0.50%
      Total rx-rate: 160Mbps
      Median Jitter: 0ms
      NOTES:  Stable results, but system is at maximum load.
   
   Test 9:  Wired-to-Station:  10 stations, 16Mbps UDP stream, 1472 byte payload
        Channel 157, 5Ghz, HT-40, NO ENCRYPTION
      Median one-way latency:  2ms
      Average packet-loss: 0.53%
      Total rx-rate: 160Mbps
      Median Jitter: 1ms
      NOTES:  Stable results, disabling encryption did not increase throughput.
              This is the max load for this system in this scenario.