• No signal strength readings were available.
• Testing done with client and AP approximately 6 feet apart unless otherwise noted
• Testing was done with a Belkin F8T002 Cardbus client card in a WinXP Home Dell Inspiron 4100 laptop, unless otherwise noted
• Response time measurement was done via standard ping test, 25 repetitions
Bluetooth is not intended to be a high-bandwidth wireless networking technology. But since I need to say something about performance, I decided to do some basic throughput testing. Because of the way the AP works, I had to struggle a bit to use my normal performance test tools, but finally managed to get a Chariot run done, with the result shown in Figure 5.
Figure 5: Bluetooth throughput
As expected, throughput isn't a barn-burner, averaging about 180 kbps. This test was run without using encryption, but enabling it and rerunning the test didn't show any appreciable difference in connection speed.
Belkin supplied me with three Bluetooth client cards for my testing. The above plot was done using a Cardbus (!) based F8T002, Class 2 (10M range) device, but I also had two USB-based adapters, one Class 2 (an F8T003) and the other a Class 1 (a 100M range F8T001).
Since even USB 1.1's 12Mbps max raw bit rate far exceeds Bluetooth's highest asynchronous raw data rate of 723 kbps, I didn't expect to see any difference in throughput. But Figure 6 shows a different story!
Figure 6: Bluetooth throughput - Cardbus vs. USB clients
(click on the image for a full-sized view)
This plot compares Chariot runs using the CardBus based F8T002 and USB-based F8T001 and clearly shows a 30% lower throughput from the USB-based adapter! I have no explanation as to why the USB adapter should have lower throughput than the CardBus adapter, since both bus types have more than enough bandwidth to support Bluetooth's needs, but repeated tests duplicated the results.
I also started to run throughput checks at my normal test locations, but quickly discovered that the range was pretty much limited to inside my office with the Class 2 device. I could still get a very slow connection when I moved my test notebook about 15 feet away into the hallway that leads to my office (with my office door open). But putting one drywall-type wall between my notebook and the AP completely killed the connection. So even though the AP is a Class 1 device, you don't reap any benefit unless you use a Class 1 client with it because the weaker Class 2 signal determines the operating range.
Substituting the Class 1 USB adapter, however, considerably improved the range. I could move down the hallway about 25 feet from the AP, and still get essentially the same speed. But ducking into the room at the end of the hallway - putting two walls between the client and AP - quickly killed the connection. I was also successful in getting a top-speed connection directly over the AP one floor above. So although you'll pay a speed penalty by using this adapter (which also has a cute little flip-up antenna), you will get improved range.
Bluetooth Wireless Performance Test Results
|Test Description||SNR (dB)||Transfer Rate (Mbps)||Response Time (msec)||UDP stream|
|Throughput (kbps)||Lost data (%)|
|Client to AP - Condition 1||0||0.1830||44 (avg)