wiki:802.11/Benchmarks

Version 13 (modified by murphpo, 10 years ago) (diff)

--

The 802.11 Reference Design and its documentation are under active development by the Mango team. The current release should be considered a beta- updates with bug fixes, API changes, new features and other refinements will be posted frequently. Please check the downloads page for the latest updates and post any questions about the design to the forums.

802.11 Reference Design: Benchmarks

Placeholder page for eventual library of benchmarks of non-beta 802.11 ref design.

Throughput vs. Payload Length

This test measures the achieved throughput of each PHY rate between two WARP v3 nodes running the 802.11 Reference Design. One node acts as AP, the other as a STA. Traffic is generated locally via the LTG framework; measurements are extracted from the nodes via WARPnet. The nodes' RF interfaces are connected via a coax cable and 50dB attenuation.

The plot below clearly illustrates increasing throughput with higher PHY rates and the decreasing impact of MAC overhead with increasing payload lengths.

Download PNG

Competing Flows

This test measures the achieved throughput of two competing flows between two nodes. Each node sources a backlogged flow of max-length UDP packets for the other node. The contention between flows is managed by the 802.11 DCF.

Traffic for this experiment is generated using iperf on OS X machines connected to each node via 1Gb Ethernet.

The chart below shows this test for two hardware setups:

  • W3-to-W3: Two WARP v3 nodes each running the 802.11 Reference Design, one acting as AP, the other as a station
  • W3-to-WRT: One WARP v3 node connected to a Linksys WRT54G node. The WARP node acts as AP, the WRT54G as a station

In both configurations the devices' RF interfaces are connected by a coax cable and 50dB attenuation.

Download PNG