Getting Ready to Test
Broadcom's key allegation about Super-G is that it will severely interfere or, in some cases, completely shut down nearby non Super-G 2.4GHz-based wireless LANs, i.e. the "bad neighbor" effect. So I assembled two wireless LANs, configured into the test bed shown in Figure 8.
Figure 8: Basic Super-G Test setup (click on the image for a full-sized view)
The setup has two completely independent wireless LANs - one Super-G based and the other using gear based on Broadcom's 11g chipset. All APs and Ethernet test partners were plugged into a 10/100 switch and everything was assigned IP addresses in a single Class C subnet (192.168.3.X).
Since one of the two WLANs had to move to test at varying distances, and because I have only two notebooks, I used them both in the Super-G half of my test rig to make things easy to move around. This meant using a desktop machine and 11g USB2.0 adapters in the 11g half of my testbed, which remained stationary for all tests.
Tip: As I was putting together the testbed, I found Win98SE to be limiting the Super-G throughput results. When I switched the same computer from running 98SE to WinXP Home, Super-G average throughput (with all speed enhancements running) jumped from 29Mbps to around 37Mbps.
Ixia's IxChariot was used for all testing, with the console running on the Super-G laptop.