2004 WiFi Shootout

The adventure began in 2003. We wanted to wirelessly network, using 2.4GHz 802.11b, our houses for high speed data transfer and internet sharing. We looked at the topography and thought that it "might" be possible. The problem is that Justin lives about 20 miles away from the nearest one of us (Ben and Andy, who are more centrally located), and Brandon lives about 15 miles south. All of us are located relatively high in relation to the surrounding land, but not above the trees,which would probably pose the biggest problem. We starting collecting parabolic dish antennas at the 2003 Dayton Hamvention in May. We picked up a couple in the three foot range or so.

Then we started thinking bigger and realized that large unused satellite dishes were all over! So, we went around Ben's neighborhood asking people with large dishes if they were using them - with only one or two exceptions, they were not, and were willing to let us remove the eyesore from their yards for free. It was a win-win situation. We picked up four large dishes this way. We would usually load up Ben's Jeep and Andy's truck with tools, and park the truck under the dish. Many times we would just Sawzall the post and it would fall into the truck. Of course, loading and driving with a 9.5 foot dish in a 6 foot wide truck is interesting, but most of the dishes were within a couple miles of Ben's house - one was actually about 30' from his back fence so we just tossed it over.

After thinking about it and looking at topographical maps some more, we decided that we would really need to get the dishes pretty high to connect, probably about 50 feet would have done it. However, Andy was leaving home in the fall to start college and the rest of the guys were doing the same the next year, so it would have been a lot of work for a short period of use. And of course, our parents were not all for us hanging 7.5 to 9.5 foot dishes many feet up in the air on their property. What eventually happened is that the dishes sat around unused at Ben's place.

Fast forward to the next summer. We were again thinking about wireless networking, but on a much smaller scale. Ben had just bought a USB Orinoco card and needed a use for it beyond the typical home wireless network (that's nothing new). Recognizing through wardriving how many unsecured access points were around, we wanted to help the owners of these networks and make them aware of their vulnerability. If they wanted, they could let us secure them for a small fee. Ben made some directional antennas based on plans available on the Internet, and Andy and Ben set off one evening in July in search of unsecured networks. We indentified probably around ten in a couple hours time. What took time was locating them to one house. We had flyers printed up that we filled in the SSID and IP of the network, and handed these out. Most of the time when someone was home they met us with blank stares and looked somewhat freaked out that we knew these things. We returned to Ben's house after our fruitless efforts (no one called back later either) and started looking online for higher gain antennas. Andy happened to see a website about the Defcon Wi-Fi Shootout 802.11b distance competition near Las Vegas. Filled with overconfidence and youthful hope, it was then that Andy announced to Ben that they were going to Vegas.

"When?"

"In 19 days."

"Riiiiggghhhhtttt."

Among this group, this type of half-joking, half-serious conversation is the norm, but this seemed like a pretty lofty goal at the time. Of course, Andy was thinking about the 9 foot dishes that were being overgrown with brush at the edge of the woods. Ben was quickly won over to the idea, and we set out with flashlights at about 11PM to take another look at the dishes. We pulled out two of the more promising ones - two 9.5" steel dishes - both the same model (Beta 9's). They both were composed of eight pie-shaped reflector sections bolted to a back hub or "doughnut." This meant they could be unbolted for easier transportation across the country - a feature that most aluminum or fiberglass dishes don't offer.

We decided that since the competition was less than two weeks away, we could work at it and if we didn't get something together, it wasn't too much effort wasted. We called up Justin, who was a bit harder to convince. He and Brandon were in by the following day.

We had a bit of work ahead of us - we needed to build feed point antennas for the dishes ("feeds"), mounts for the mobile dish and the base camp dish, and of course, perform testing of the system before we left. We sort of wasted the first real night (we were working normal jobs during the day) on the project looking for locations to test. Andy had purchased TopoUSA - a topographical mapping program by DeLorme, at the 2004 Dayton Hamvention, that could plot the topographical profile along a straight line between two points, so it was very useful in determining places where we could make line-of-sight links. At higher frequencies, such as the 2.4GHz band that 802.11b operates on, both ends of the wireless link need to be "see" each other at the longer ranges that we were trying to accomplish. Obstacles in the line of the signal cause a lot of attenuation - even trees and foliage are a problem this high in frequency. The first night we went to a location that looked like it might work - it was about 6 miles apart, with one end in a cow pasture in Kentucky looking down over the Ohio river and I-275 bridge, and the other end in Indiana - it was basically looking up the Great Miami River valley. We thought that by shining spotlights at each other across here we would be able to make sure we had line of sight for later testing. It turned out to be unsuccessful.

We turned our attention to the feeds for the antenna. A parabolic dish antenna (most commonly used for satellite communications) works like a flashlight or car headlight. A radiating element (the bulb in the flashlight analogy) sends a signal towards a parabolic reflector. The reflector is shaped in a way that it gathers the energy coming from the point source (filament of a flashlight bulb) and reflects it in a single direction. The same is true for receiving with an antenna - the reflector gathers signals from a larger area and focuses them at the focal point, resulting in a much stronger signal. The focal point is very important - if the feed is not at the correct length and point (which for our dishes is centered - smaller satellite TV dishes are offset designs - the feed is still at the focal point, but it is not centered on the dish) antenna efficiency is much less. Also important when using parabolic dish antennas is the directionality of the feed - if it is not matched to the shape of the dish the gain (effective amplification of transmit and receive signals) will be lower. If the feed is too directional, only the middle of the dish will be "illuminated" by the feed, and the larger size of the dish will be for nothing except added weight, wind load, and transportation and handling difficulty. If the feed is not directional enough (emits over too wide of an angle) much of the signal will shoot past the edges of the dish and be wasted.

There is an intrinsic characteristic of parabolic dish antennas called the f/d ratio (focal length divided by diameter). This is what determines how directional the feed must be. The f/d of our antenna was about 0.30 (unitless since both numbers are length), which means that the focal length is short compared to the diameter of the antenna, so the feed must have relatively low directionality. Modern satellite TV receive dishes have higher f/d (around 0.6 or so) and are flatter, while dishes with low f/d (like those we used) are more curved ("deep dishes"). The feeds we originally intended to use were the biquad design popularized by Trevor Marshall, which are the most commonly used when converting small satellite dishes to 2.4GHz operation. However, after learning about the f/d characteristic, we realized that were too directional for our application, so we searched for a more applicable design.

We mainly found ham web sites - they seem to have the most public information on larger parabolic dish antennas. We found a lot of good information on the web page of find call. He has a number of patch-type feeds described that work well with lower f/d dishes. We built two feeds based on his 2.4GHz patch description (conveniently, hams are designated by the FCC as the primary users on the low end of the 802.11b band so lots of ham-generated information applies to 802.11b as well). Patch antennas are often used for GPS receivers, because they are thin, but their radiation (and receiving) pattern covers much of the sky. We built the antennas out of brass from a hobby store, because it is easy to cut and solder connectors to. After building them and mounting N connectors, I tuned them using a network analyzer at work - the frequency response was centered right about where we wanted.

We decided to use the USB Orinoco cards for the radio part of the system - they are inexpensive (<$40 on eBay when we bought them) and offer some great benefits - they have a connector to bypass the internal antenna, so it was easy to connect the feed without modifying the cards, and they are USB, so we were able to mount them only a few inches behind the feed. Instead of running expensive low-loss coaxial cable between the feed and the card (which is typically installed in a laptop), we converted the signal from RF to data only inches from the feed (only two connector pairs) which resulted in very little RF line loss. We just ran a USB cable down the tube holding the feed.

The feed assemblies were made from PVC pipe and turned out quite well, we think. A 1" or 1 1/4" pipe mounted in the center piece of the dish with existing set screws allowed us to easily adjust the distance of the feed so we could get it right at the focal point. This small pipe had a large bulb looking piece on the end made from about a foot of 4" PVC pipe with three end caps - one on each end and a modified one inside the end nearest the dish. We drilled a hole in the two end caps nearest the dish to mount to the smaller pipe holding the assembly in place. The feed was then trimmed to fit between the end of the pipe and the far end cap, with a hole drilled for the Type-N connector. Originally we only had one Orinoco cards that we mounted on the end of one of the feed assemblies, we ran a thick coax cable from the other feed assembly to a modified access point for testing while the second Orinoco card was in the mail.

Dishes as large as ours require a bit of effort to move and position. Fortunately, we have access to quite a bit of metal and between the four of us we have about five welders of various types. We made the first mount on the first weekend into the project (only one of two we had to work). We pulled out one of the dishes and leaned it up against a scaffold we made. First, we needed to come up with a method to position them once they were mounted. The original idea was to use two lengths of all-thread rod (one for azimuth adjustment and one for elevation adjustment) and a nut to slide along the all-thread as it was spun. We used two casters to let the ends move how they wanted (the angle of the all-thread with respect to the mounts changes as the moving nut travels along the length) - one was welded to the moving nut, and the other was welded to a pipe sleeve that let the all-thread rotate in it (the fixed end). We mounted one caster to the existing pole that was originally used to mount the dish in the ground. The other caster was mounted to the arm the pivoted the dish. The other actuator set was mounted similarly to the other adjustment axis. After getting this set up, we used an impact wrench to turn the all-thread. This was a pretty crude system. One of the moving nuts somehow got cross-threaded in the middle of the all-thread. Then they other all-thread started bending. We decided we needed something better.

We soon thought of a much better solution - most of the dishes we picked up had "dish mover" linear actuators. These were great - because they have DC motors (easily reversable) that will run on 12 volts, and they are designed to work with the large dishes. We only had two working ones, which we mounted on the mobile dish. Once we got this working, we wired up a switch box with two DPDT (double pole double throw) center-off switches - so switch it up to move the dish one way, and down to move it the other. This was very nice. We then made a frame to mount the dish in the back of Andy's truck. The original - in ground mounting post was mounted horizontally, using landscaping timbers and 2x4's, above the bed. With the dish in it's operational position, it pointed straight out from the back of the bed. The problem was that it was over nine feet wide - we needed a way we could turn it sideways to drive it around. With the linear actuators, it could turn far enough to the side to be within the width of the bed, but it couldn't turn this way because it was too low and would hit the side of the bed when it turned. This problem was solved by putting a jack under the back of the mounting frame. The back was jacked up when stowing or unstowing the dish for driving, and let down before driving or aiming the dish. This method worked out pretty well.

We did some initial testing once this dish was mounted, but couldn't get good results. We tried aiming it at a laptop about 100 yards away, but didn't see the big signal gains we were expecting. We still had another dish to mount, so we worked on that the next week. We had some triangular tower sections that Andy had gotten about four years before. After sitting in Andy's yard for a few years, they made their way to Justin's house where they sat for a couple more years. We pulled them out and came up with a tripod arrangement to mount the dish. We got the center of the dish up about 10 or 12 feet this way, and it was pretty stable. With the tower sections, it was easy to climb up the back if needed (or just for fun...). We first tried using U-bolts to hold the vertical sections to the more horizontal legs, but these bent quickly under the load of us climbing up.

What we did instead was weld steel plates to the sections where they joined, and bolted the plates together. Arc welding a couple feet of bead in the rain was interesting. This method was very strong (bolts in shear instead of in bending like with the U-bolts) and easy to set up. The only issue was that we didn't have any more working linear actuators. We were planning to test the Sunday before we would have to leave, and it was about Friday by this point. On Saturday we were still short two linear actuators, so we canvassed our neighborhoods and were able to come up with two more dishes (to be collected later), but more importantly, working actuators. We installed those and got everything ready for testing on Sunday.

After returning from church on Sunday morning, we hauled everything out to a St. Xavier HS baseball field parking lot overlooking a valley. We thought it would be easy to get multiple mile line of sight links there. We set up the base mount dish, hauled down from Justin's house, and did the first test about a hundred yards away. The signal levels were higher than we had seen before: -17dBm. A wireless link will work down to about -80 or -90dBm, a typical home link is around -60 or -70dBm (a 10dB difference is a factor of 10, -60dBm is 10 times stronger than -70dBm). We were encouraged by this and set off to make a longer link. We could see many places, but it was hard to determine where they were in terms of how to get there. We found a good starter location - just across I-75 in an industrial area, about 2.5 miles away. We could physically see the other antenna from each end, which was good. We made this link at about -60dBm and adjusted the feed length on both dishes. One cool thing that we started with this link was doing a remote desktop connect to the far computer. We headed back up in the truck and looked for another location (and ate lunch). We didn't really see any good locations, and it was getting late, so we set off in the truck again and tried to find some good locations. The difficult part was correlating what we saw at the base location to actual places. We ended up not making another link after driving for over an hour, so we headed back as it was getting dark and beginning to rain. We loaded up the base dish in the rain and left.

An interesting thing from this test was that Andy's dad, an RF engineer, saw the setup for the first time and was pretty impressed - he had heard plenty about it but apparently wasn't expecting much. He realized that they actually had a shot at the contest. Our parents called a meeting for the following night at Justin's house, since that was where we were planning to test. They wanted to discuss little things like how we were going to get there and where we were going to stay. At this point we didn't know what we were driving and were planning to camp in the desert.

The mobile dish was switched from Andy's truck to the one Brandon had (his parent's), because Andy had to work the following day while the others went out testing again. We decided to test near Justin's house because there was about a nine mile link with one end only about half a mile from Justin's house, with the other end in Hamilton.

The next day at Justin's house we asked a farmer if we could set up in his field, which he let us do. Andy was at work, where he talked to a guy who said that we should contact his brother-in-law, who owned a local company, RF Linx, that made Wi-Fi amplifiers! Up until then, we had been planning to run in the unamplified class, but amplifiers would allow us to run both amplified and non-amplified (by taking out the amps). So we called up RF Linx and borrowed the amps. With these, we were able to make the nine mile link in the rain, which we hadn't been able to without the amps.

Everyone convened on Justin's house, and Andy brought the second Orinoco card. First, we hauled everything back to Justin's house from the field. We then had the meeting with our parents, which was interesting. Basically, they told us that while technically we were prepared, we hadn't planned the actual trip enough. They tried to talk us out of it and even offered an attractive alternative (boating in Michigan for the weekend). Also, at that point we were not planning to have a parent go, and our parents were uncomfortable about letting their three teenagers go to Las Vegas. Brandon wasn't able to go at all, due to a pre-planned family vacation. Justin was not allowed to go without a parent going. However, we discussed it amongst ourselves and decided with the work we had done, we wanted to put in a couple days of hard work in terms of finishing getting ready and packing everything. Justin's dad was willing to take off work and drive their van and trailer out with us. This worked out really well and we are glad he was willing to do that for us. He also made hotel reservations so we had somewhere to go when we got there!

After the meeting, we got the dishes apart - we had to use a couple impact wrenches to get the rusted bolts out. The next day we still had a lot to do - get hardware to put the dishes back together, get printed topographical maps, and the many other little details that always come up before a trip. Brandon took care of a lot of stuff for us that day, which was a big help. We did much of the same Wednesday morning, including outfitting the van with radios and computers so we could bear the trip, and left around 3 or 4 PM that afternoon.

We drove straight through, trading off between drivers. We made it to Denver about noon on Thursday, and from then on the drive was pretty interesting because the scenery is so different from what we are used to. It was neat to be at about 11,000 feet crossing the Rockies and seeing snow in the middle of summer! We also did APRS (Automatic Position Reporting System) using Justin's Kenwood TM-D700 so that our families could keep track of our progress on the Internet. We also used Justin's Garmin iQue (PDA with GPS) for voice navigation throughout the trip. Working at a radio store (R&L Electronics in Hamilton, OH) has it's benefits!

We arrived in Las Vegas, NV after about 34 hours. Our hotel was on the north end of town, and we crashed when we got there (around 11PM).

The next morning we went to Home Depot for some supplies (struts to brace the base tower assembly, step ladder to get up to the feed to switch the amp in and out, etc), ate, and then headed down to the Alexis Park Hotel, the site of DefCon, and looked around a bit. We met up with the WiFi contest organizers, registered, and found out the location of the contest. It turns out it was being held at the same location as the previous year. We headed up I-15 and then up 93 to the base camp site. The valley we were shooting our signal north through was pretty impressive - we can't see that far anywhere back home! We could see about 20 miles, about the distance from Ben or Andy's house to Justin's - that's a long way to be able to see!

We met the judging team - ASLRulz- they were the winners of the contest the year before, with a 35 mile link using a horn antenna built from conduit and chicken wire. They had done a lot of preparation for the contest. They had flown over 10 people out from New York, rented an H2 and other vehicles, and set up a radio link for themselves and the contestants to use. They were a group of hams, like us, and the radio link consisted of a 2-meter (146MHz)/70cm (440MHz) crossband repeater at base camp. They had a 2m yagi antenna pointing up the valley for the mobile teams to use, while the teams at base camp used the 70cm channel. These were simultaneously linked together so the teams could communicate between base camp and the mobile team. Cell phone reception was spotty, so some of the non-ham teams had difficulty communicating with their mobile station. We had brought a large 2m yago ourselves that we planned to use, but it didn't work for some reason so we used the provided crossband repeater. It worked very well, although even with the higher gain antenna we could only communicate at the longer ranges when the van was high enough above surrounding hills. The judging team took the judging seriously, but was also fun to work with, which made a big difference.

We began the task of unloading and went up the hill to look for a good place to set up. We didn't want to have to go too far because carrying several hundred pounds of dish, mount, batteries, etc up a hill gets tiring pretty fast, and it was made worse by the 109F or so temperature. We found a spot on the ridge of the hill that looked like a good place to set up. We began the task of hauling stuff up the hill - the prickers that get in your shoes really suck! We got everything in place to start testing on Friday around 4 or 5PM, and Justin and his dad drove out about 14 miles with Ben and Andy at the base camp site. We asked for an official distance measurement, which we met.

The judging method was comprised of two elements - a group at base camp and another couple guys who would drive out to the location of the remote team. Each of the groups had a GPS receiver to get the positions of the antennas, and then they would provide a three word phrase that we would send to the other team. Actually, what we did was just save the phrases in a text file, and then the other team would use file sharing across the wireless link to access the text file and report the phrase to the judges on their end. The process would then repeat in the other direction, and the judges would acknowledge it. That was to satisfy the judges, but we also did remote desktop connect between the two computers so we could actually see the remote screen on ours in order to help align the antennas (we would run the Orinoco utility that let us see the signal levels on both ends). That was pretty cool.

We drove out to 25 miles to make sure we could make a link that far, which we did without a problem, but the judges had left by that point (it was around 7PM or so). We drove back to base camp, although we had an interesting problem on the way. The azimuth actuator was held on one end by an adjustable sleeve. We would swing the dish to the side to minimize drag while driving and to stay within the width of the van/trailer (instead of 9.5 feet wide). Apparently this sleeve wasn't tight enough, because the actuator slid in it and the dish went straight back and acted as a parachute, slowing us down pretty quickly. We loosened the sleeve, slid it back and position, tightened it down (really well) and it wasn't a problem again.

We unloaded the mobile dish from the trailer, drove back into town, ate, and went back to the hotel. We got up the next morning and headed back out to base camp around 10AM. We loaded the mobile dish back up, and Justin and his dad took off in the van while Ben and Andy stayed on the hill again. We were ready to start getting official measurements shortly after noon, which is when the judging began. We made the first judging request at 25 miles out, near where we had tested the night before. We proceeded to work our way out the rest of the day. After checking the topographical data and elevation profile using TopoUSA, we saw a spot about 38 miles out, several miles up a hill on a gravel road. We drove to this location and set up. When the judges arrived, they said that it was the same road that had produced the winning distance the year before - but they only had a car and were not able to get as far as we did with the van (towing a trailer).

We did two tests at each distance - one amplified and one unamplified. We did the amplified test first to make sure that we had good line of sight and antenna alignment. Once this was successfully judged, we took the amplifiers out of the system. We actually physically removed the amplifiers from the system (there was no provision in the amplifiers for bypassing them, and it visually showed the judges and anyone else that the amplifiers were not used) - this involved pointing the dishes down so that we could access the feed, pulling the feed apart (PVC pipe - friction fit instead of glued), removing the amplifier, reconnecting the feed, and putting the feed assembly (pipe) back together. Then we raised the dish back up and got close to the previous aiming. While not extremely precise, we didn't have problems with this method - we also monitored the signal strength and fine tuned the aiming once we were connected.

When the distance was confirmed by the judges, there was talk that this might be an unamplified world record. This was exciting, but we still had some time left and thought that if we could find a more distant location, we could make an unamplified connection even farther. The only problem was that we didn't know where to go next. After some frustration while searching the topo data (there was no way to tell the condition of the indicated roads - except for the main highway, almost all were gravel or just cleared ground), we found a location that we thought "might" work. There was only about an hour left before the 6PM deadline, so we were cutting it close. We saw a road indicated on the map as the same as the one the van was currently up. The only exception was that instead of a few miles to a clear LOS site, it was about 15-20. We didn't known if the road was in decent shape all the way up, but we had to give it a shot. Further compounding the situation was the fact that at the longer distances we were out of range of 2-meter radio (or cell phone) communications except when the van was high enough. When the van left the 38 mile link site, we would not have a way of contacting them unless the van made it up to the second location. Finally, at about 5:50PM, we re-established communication and they had made it up! We were really pushing the deadline at this point! We had to get an amplified confirmation, remove amps, and then re-establish the link with no amplifiers. We successfully tested the amplified link, and hurriedly switched over for the final test.

We estimated the antenna alignment, and nervously checked to see if we could see the far station. We had it and were able to connect! However, we had very low signal strength and didn't even know if the link would remain with slight wind movement on the dishes and other minor factors. It was about 5:58 or 5:59 by this time, and we didn't even have time to align the antennas as we had before. We quickly asked the judges for the official test phrases, and the link stayed up throughout the transfer, much to our relief! The judges confirmed our unamplified connection at 55.1 miles and we were ecstatic! However, we were also pretty worn out from working in the 108F or so heat all day, so we began to pack up.

By the time Justin and his dad arrived back at base camp, Ben and Andy had hauled quite a bit of the equipment back down the hill. We talked to the judging team as we were dismantling the setup and loading up, and they thought that we had set an unamplified world record. This was pretty exciting - we had come out to win the contest without much thought toward actually setting records, so that would be a nice bonus! We called Brandon and our parents on the way back into town - they were all pretty excited too! We actually had a bit more time on Saturday night, so we dropped off the trailer at the hotel and went down to the Strip. We saw some of the sights in downtown Vegas, and drove the length of the Strip - it was cool but we were pretty worn out and were trying to stay awake! We had dinner and went back to the hotel.

We got to sleep in a bit more Sunday, and headed down to DefCon in the afternoon. We wanted to take some of it in - the meetings we sat through were cool - we missed a lot while we were out in the desert! The awards ceremony was the last official event, and we were excited about it. It lasted for quite a while as other contest winners went up. By the time the WiFi Shootout judges went up to report on the event, we were getting excited. We were announced and went up. The distances were announced incrementally, and we were surprised by the big audience reaction to the lower distances - I think the buildup was more exciting for us than for the audience! We didn't know what the would do! When the 55 mile unamplified distance was announced, we got a standing ovation from the crowd of several thousand! We were given the opportunity to talk to the audience for a few minutes, which was exciting. I think Ben gave the entire life story of the project. We were well rewarded for our effort - many books about wireless networking, wireless hardware, and best of all - a total of $1500 in Best Buy gift certificates!

After the ceremony, we tried to see if anyone was interested in buying our equipment, or even taking it for free! Justin's dad didn't want to haul hundreds of pounds of steel back across the Rockies when we had several more big dishes laying around our yards at home (and apparently the thought of two more dishes rusting away wasn't appealing...). We ended up leaving it in the dumpset of the hotel before we left town.

We took a much more leisurely trip on the way home - first stopping at Hoover Dam. As everyone has said, it is amazing - you can't really understand the scale of it, even when you are standing at the top looking down. We spent the night in Zion National Park, and explored there the next day. Ben got an interesting call as we were leaving - the Cincinnati Enquirer (the local newspaper) wanted to know about what we had done! Apparently they had seen a write-up on the Wired web site and got our contact information. Ben talked to them for a while and we were excited about the possibility of a newspaper article. We spent the next night outside of Arches National Park. The next morning we drove around the park a bit and headed to the Rockies to spend the night in Rocky Mountain National Park. We drove over Trail Ridge Road (came into Estes Park the back way) - that was high (goes over 12,000 feet). We spent the night in the campground and drove back up the mountain the next day. We had a hard enough time walking around - it was incredible to see people cycling all the way up at those altitudes! We went back down and began the drive home. We drove straight through from there, and arrived home on Friday afternoon.