How we provided WiFi at Maker Faire: a detailed explanation

November 1, 2016

At Meta Mesh, we build Community Wireless Networks. But these skills can also be used for other types of setups. Today I'll be talking about building an event-wide wireless deployment.


The Setup:


The organizers of Maker Faire Pittsburgh got in touch with us when they needed an emergency event-wide wireless network for their Vendor tents and Ticketing Booths. The purpose of this network would be to allow Vendors point-of-sale and primarily allow volunteers to sell tickets on iPads/laptops and check-in members. Below is the area we were asked to light up.




We did not have enough time to order new radios for this event, so we were left to use what we had on hand and in spare radios/antennas. This forced us to get a little creative with providing a reliable solution!


The Backhaul:


There are two sets of frequencies predominantly used for WiFi, the 5GHz and 2.4GHz bands. In an ideal setup, we would have used 5GHz for every leg of our impromptu network as there is more spectrum to work with in 5GHz and more options to avoid interference. We were tasked with covering the Buhl Park area of the event space. This space is sandwiched between the Children's Museum, an office building, and apartment buildings which meant the 2.4GHz Wifi spectrum was extremely crowded.





One of the ways we got around this interference was to use Dynamic Frequency Selection channels. See, in the 5GHz range, there is a particular portion of the band that is shared between WiFi devices and a kind of weather radar. To ensure that WiFi devices don't interfere with your five-day forecast, WiFi radios need to be able to listen for pings from the radar stations and quickly jump to a different channel. The FCC is concerned about this interference and it was this problem that inclined them to attempt to sort-of-but-not-really-but-actually-kind-of ban OpenWRT, DDWRT, and other third party firmware from being installed on routers. The devices we decided to use can broadcast on these channels- but a lot of consumer-grade devices don't because that's one less thing for manufacturers to have to deal with when selling their products in the US.


So for our backhaul to our Internet gateway, we used two 5GHz Ubiquiti Rockets. We used DFS channels, moving us into a very wide open range of the 5GHz spectrum with less interference.  


We still had a fair amount of physical obstruction though: a bunch of trees, a tent, and, well, people which we could not work around. Ideally we would want to use a WiFi technology that bands two channels together for max bandwidth- but using two channels means that twice as many other devices would interfere- so we narrowed our channel use to 20MHz. We were able to achieve better link quality overall because of this and we were able to get our backhaul link to about 35/35Mbps, which exceeded our Internet connection of 15/15mbps.


The Access Points:


So we had a fast-enough connection back to the Internet. But what about the WiFi that people's phones and laptops connected to? The next stage of the setup was a single 2.4GHz Rocket in a point-to-multipoint setup. We used a 120° antenna allowing us to use two 2.4GHz Rockets pointed in different directions at the vendor's tents. This was the trickiest part of the setup, as the 2.4GHz spectrum in this area was MASSIVELY crowded. At 20MHz channel width, the only non-overlapping channels in the 2.4GHz spectrum are channels 1, 6, and 11. The image below shows a spectrum analysis of the area:




You can see the three "humps" in that graph showing how 1,6, and 11 are just being brutalized. Throughput speeds were approximately 5mbps; slower than Pigeon-Net.


If we stayed at channel width of 20MHz, our 2.4GHz links would be extremely unreliable and slow no matter which channel we selected. However, take notice of the peaks and valleys on the middle graph. What we decided to do was to shrink our channel width and sneak into a dip in the spectrum!


The decision to do this greatly improved our link quality: re-transmitted packets (a bad thing) went from 30% to 1-3%, AirMax Quality, a measure of connection quality, doubled and link capacity tripled. We eventually settled on 8MHz channel width at Channel 9, and while we did sacrifice total bandwidth because of the smaller channel size, it was worth it to reduce interference and improve reliability of the 2.4GHz links.



Remember, bandwidth out to the Internet was 15/15mbps, our backhaul link was 35/35mbps, and we got each of our two 2.4GHz links to do ~16/16mbps. Huge bandwidth was not a requirement for this setup. Our goal here was to provide reliable Internet to the Vendors and Ticketing Stations at the Maker Faire, making reliability of the links more important than pure speed.


The last piece of the setup was the Access Points (AP) to which the Vendor and Ticket Station phones, tablets, and computers would connect. The two 2.4GHz links described in the previous section were used to feed these two Access Points (AP-A & AP-B). Unfortunately we had no dual-band (2.4+5GHz) access points on hand, which would have been ideal.


Two Access Points (AP) meant we had to use channels 1, 6, or 11. Since our 2.4GHz AirMax Links were on Channel 9, using Channel 1 for AP-A was an easy choice, but left 6 or 11 to use for AP-B. Since our 2.4GHz Link used Channel 9, no matter what we chose, we would be interfering. We decided on Channel 6 since it was the furthest from 9.


We also used this channel on AP-B the most “out of the way” from our Link. We used a long internet cable to run to this AP-B, so it was a bit more out of the way than AP-A. This didn’t eliminate interference completely, but it was the best we could do to reduce it.


The picture illustrates this best (not to scale):




This network served about 35 people over the 3 days it was operational and passed just a bit over 3GB of traffic. These aren’t sexy numbers, but as we mentioned, the purpose of this network was to provide reliable connectivity for Faire operations. It accomplished this despite a very hard to work with 2.4GHz environment and a bit of a time crunch. I could go into a lot more detail, but this is more of an introductory post. If you’re interested in learning more, get in touch with us! We love talking about the work we do!


Overall it was a resounding success. We achieved better speeds than the "professionally installed" WiFi that already served the other side of the Faire. We're happy to have it come together so well and can't wait to do it again at next year's Maker Faire!



Share on Facebook
Share on Twitter
Please reload

Featured Posts

State Of The Mesh: March 2017

March 2, 2017

Please reload

Recent Posts
Please reload