First-ever 802.11 rocket telemetry

Recently, we at PSAS are likely to have been the first ever to successfully use 802.11 as a rocket telemetry system. I posted an article about this to /., but was rejected. So I'll post here, and hope that folks are duly impressed…

The pieces we used to make an 802.11 link work at above Mach 2 at 3.5 miles up were manifold.

• 802.11 insists on doing a bunch of handshaking below the link layer to try to avoid collisions and improve transmission reliability. This is very problematic for a telemetry system, which needs to have full control of the transceiver at the application layer.

  Fortunately, our resident mobile wireless and network guru Jim Binkley was able to suggest a workaround. We got Orinoco cards, and ran them in this funny ancient pre-802.11 mode called "ad hoc demo mode". Broadcast packets sent in this mode are simply dumped out with no handshaking.

• Good antennas are a necessity for reasonable performance. On the ground side, Steve designed the Trackmaster 2000
  

  

  to aim a bunch of quarter-wave helicals designed by Glenn at the rocket. These included an 802.11 antenna.

  On the rocket side, the team spent 2 years designing, implementing, and testing cylindrical patch antennas.
  

  

  We had offers to make these antennas for $15,000-$18,000 apiece. The end product of the work is an open source spreadsheet that produces Gerber PCB layout for a given antenna. The resulting antennas can be fabbed for about $100 each.

• Even with antennas, good amplification is a must. We had 1W amplifiers on both the rocket and the ground side. Since we were operating on 802.11 channel 1, we were operating in the amateur radio spectrum, so were legally licensed to do this.

In its one test flight, the system worked spectacularly well. In several minutes of operation from launch to near the ground, only three packets were lost. Since the vehicle itself was utterly destroyed, the resulting recorded telemetry downlink is the only flight data retrieved from the rocket.

We are utterly happy that this worked so well—it represents the culmination of a ton of hard thinking and implementing. The number of people who told us plausible reasons why the system "couldn't work" was large, making the success that much more satisfying. (For example, the Doppler shift at Mach 2.4 is only a few kHz, far less than the drift/wobble on the GHz oscillator in the transmitter and receiver.)

That said, I think we're going to try for a custom spread-spectrum telemetry link for the next rocket. We've made our point, and our work with GNU software radio should allow us to complete and develop the prototype system begun by Sarah Bailey.

Modern cheap COTS technology is amazing. Don't write it off as just toys, or just good in its intended appliation domain. Our 802.11 system is a good example of how you might move beyond this to exploit the full potential of COTS devices.