Visit Your Local PBS Station PBS Home PBS Home Programs A-Z TV Schedules Watch Video Donate Shop PBS Search PBS
I, Cringely - The Survival of the Nerdiest with Robert X. Cringely
Search I,Cringely:

The Pulpit
The Pulpit

<< [ Open and Shut ]   |  WhyFlying  |   [ The Big Picture ] >>

Weekly Column

WhyFlying: It's Pretty Darned Close to Free Aerial Broadband

Status: [CLOSED]
By Robert X. Cringely

WiFi-in-the-sky works very well, thank you, but what I learned about it in my aerial experiments this week included some surprises.

Before we get to the details of my aerial hack attack, reader reaction to my first column on this topic warrants some mention. Though that column was predominantly about using WiFi in the sky and that was the only technology I proposed to test, most of the mail concerned use of mobile phones aloft. Airline pilots told me how they regularly use them to call home from 100 miles out. Non-airline non-pilots told me how turning on a phone could drop a jumbo jet through something called "inductive freeze." But mainly people worried that a time might come when everyone sitting beside them will be yacking on a phone and the noise and peace of airline flight will be lost forever.

This argument is becoming moot. I think the advent of cellphones aloft is only a matter of time since several airlines now have systems in testing. So enjoy the "quiet" while you can.

Blame it on 9/11, but there is among some readers a substantial fear of authority. Even though testing 802.11 networking in my own airplane linking to my own base station using FCC-compliant equipment was not illegal in any fashion, many readers were appalled that I would even try, fearing that all pilots would be made to suffer for my foolishness. Sometimes authorities do spiteful things in over-reaction, but that's no reason to be afraid.

But that doesn't mean I'm stupid. So before heading aloft, I called Dan Schwinn for a reality check. Back in the 1980s, Dan started a Massachusetts networking company called Shiva, which he took public in 1994 and sold to Intel in 1998. Schwinn has since been following the aviation tradition of turning his big pile of money into a smaller pile of money by running an avionics company called Avidyne that makes some of the most sophisticated glass cockpits in the world. Back when he was at Shiva, Dan was also the first person to tell me about using a cellphone in the air.

"There is nothing wrong with using WiFi equipment in the air," said Dan, who works with the FAA and the FCC every day. "Any properly-designed avionics should be immune to 802.11 interference." So I hit the sky.

Here is what I wanted to test. I had a pair of Linksys WRT54GS 802.11g access points modified with Sveasoft firmware. One was installed on top of the mansard roof of my 1852 house in Charleston, South Carolina, and the other was installed in my funky homebuilt airplane. There were three matched pairs of antennas at each end of the link so I could test to see how antennas affected range and signal strength. In an effort to minimize the drag hit I'd get from slapping three antennas on the belly of my already-too-slow plane, I limited the tests to patch antennas that would protrude an inch or less and hopefully reside in the boundary layer of still air near the skin of the airplane and not create additional drag. I wanted to test range and data rate to my house and then later see if I could connect through any other access points, especially those that weren't rooftop mounted.

And it worked pretty well. Antenna orientation was critical since my plane is made of aluminum, so turning the belly away from my house was like putting a refrigerator in the data path. Also the directionality (is that a word?) of the antenna was a significant factor, especially at lower altitudes when I was effectively below the expected cone of directional coverage. Range was better at lower data rates like 1-2 megabits-per-second (mbps) where the sensitivity of the radio is higher because it is trying to recover fewer bits from the signal. It always amazes me how people complain about being unable to wirelessly link to their 384 kilobits-per-second (kbps) DSL line at 54 mbps when they could just turn down the speed to one mbps, increase the range, and still run at more than the rated speed of the broadband connection.

Generally speaking 802.11g is better than 802.11b even at lower speeds because it uses orthogonal frequency division multiplexing (OFDM), which is designed to operate in the presence of signal distortion and multipath. What this means is that in many cases you can get an 802.11g signal through where an 802.11b won't go. You don't necessarily need a straight line of sight when using 802.11g which will "bounce" around corners and down hallways, etc. In many cases, you will connect via 802.11g where -b won't even see a signal.

Under ideal conditions I found I could link to my home network and make a Vonage VoIP call to my Mom in Arkansas within a radius of 35 nautical miles. It might have gone further out, but I didn't test further out.

At this point in the story, a thousand nerds are ready to blame me for screwing-up a WiFi channel for metropolitan Charleston. Their concern (your concern -- you know who you are) is that by having a WiFi access point and high-gain antenna running in my airplane, I'm laying down a swath of interference and channel-hogging that will screw up someone's online gaming experience.

No, I won't.

"You are using the channel, therefore it can't be used by others (damn you)," goes the claim. That would be bad enough in one neighborhood, but since my aerial aerial can reach hundreds of neighborhoods, I'm screwing it up for thousands of people.

No, I'm not.

First, most access points allow 64 or 256 simultaneous connections. When you are checking your e-mail on WiFi channel 6 down at Starbucks does it mean that nobody else in that part of town can use channel 6? No it doesn't mean that at all, nor can it if WiFi is to be successful (as it already is) on most university campuses. It just means that you all have to share whatever bandwidth is available at Starbucks, but unless you are trying to do streaming video it probably isn't having much impact on anyone else. So if there is any impact it isn't on the channel but on the backbone bandwidth of the one particular access point to which I am connected, NOT EVERY AP IN TOWN. In the first part of my testing I was only trying to link to my own AP, so the only bandwidth I was stealing was my own.

By limiting my connection speed to one mbps, I not only increased range, I also decreased the interference I was causing other users since my signal could hop around theirs. Frequency hopping is more polite than direct sequence.

One thing that surprised me in the testing was the effect of using high gain antennas. They worked better, of course, if I could keep them aimed at the AP, but that would require either flying very high, or flying in a peculiar arc around Charleston, with my plane in a constant bank, holding the belly of the plane facing my house -- a kind of outside turn that is hard to hold steady unless, like me, you tend to fly that way even when you aren't trying to. But what really surprised me about high gain antennas is that the arc of coverage they subtended had to decrease with altitude (the gain had to go up) to maintain signal strength. So down low, I was better off with an omnidirectional antenna, but up high, a highly directional antenna was better.

Yeah, but what would it be like for people who don't like to fly around in outside arcs? What would it be like for normal people who aren't trying to connect to their own house but to any open AP? That was the second part of the test.

Several readers pointed me to a warflying account from CNN and Tom's Hardware in which two planeloads of geeks managed to find 4,000 access points in two hours of cruising over the L.A. basin. But where those guys were trying to just log as many APs as possible, I was hoping to actually do a little work. What would happen if I flew overhead at 125 mph?

Again, it varied with gain and altitude, but there were plenty of open APs for me to connect through. The problem was that since I wasn't making a herculean effort to keep my plane's belly aimed at a given AP, I had to make do with whatever my plane's belly happened to be pointing at. This changed both the nature of my connection and the way I used that connection. Those access points weren't matched twins of my WRT54GS, they didn't have high gain antennas and the antennas weren't roof mounted, so signal strengths were lower and ranges were less, with the result that I could effectively use a given AP for only a minute or two. Think about how the Tom's Hardware guys logged 4,000 APs in 120 minutes. Their average connection times were similar to mine.

So if I was going to call my Mom using VoIP using hijacked WiFly, it would have to be a short call. Effectively, VoIP was out. But everything else worked just fine -- e-mail, NexRAD radar images, anything that didn't need a continuous link for more than a minute. At first I had to adjust my behavior, but then between flights I wrote a script to manage that behavior for me, automatically looking for a wide open connection (no filters, no WEP, no passwords, just come-to-Mama) connecting and sending or receiving as needed. For e-mail I linked to my own SMTP server at home to avoid ISP mail relay blocks.

And think about the impact I had on those open access points. Whatever impact I did have lasted for at most a minute or two. I doubt that anybody even noticed.

I consider the tests to have been very successful. While I couldn't do VoIP reliably using other people's connections, I could do just about everything else. My impact on the world was minimal. And even if every airplane used this kind of networking the impact would STILL be minimal because there just aren't many planes and that number, sadly, is going down, not up in America. But best of all, if I can't call out using WiFi, then others can't call in.

Ah, blessed peace and noise!

Comments from the Tribe

Status: [CLOSED] read all comments (0)