Tales From The Towers

Chapter 58: Galactus vs. Guerilla Wireless – Our Boy is All Grown Up

By January 12, 2018 No Comments

A long time ago in a suburb far, far away, Guerilla Wireless was born. For too long, the masses had suffered with either poor or no internet or expensive single carrier options. Guerilla wireless was conceived to take over from the massive failure of poorly conceived mesh networks that failed under the weight of their over-priced, under-engineered equipment and designs. There was a hunger out there, a hunger that said, no more do we have to eat what is put in front of us or starve. Guerilla wireless fed that hunger with prices that competed with cable, at speeds exceeding DSL, and spit in the general direction of wireline. A little melodramatic maybe, but at least I didn’t mention HoHo’s.

Guerilla Wireless was built on early Ubiquiti 802.11a radios. The Ubiquiti radios were a perfect combination of inexpensive chipsets cleverly packaged to brave the great outdoors with GUI-based firmware even my mom could learn how to use. Time marched on and 802.11a became 802.11n and Ubiquiti kept making better, faster 802.11n along the same genre. Guerilla Wireless just kept getting better in the urban environment while keeping up with and surpassing DSL, but slowly falling behind cable. In the journey, Ubiquiti broke a few hearts among the 802.11n crowd by not delivering the GPS promise that would make Motorola/Cambium WISPs happy. Cambium eventually stepped in to develop 802.11n products, the ePMP line, that would sweep GPS-addicted designs off their feet, but it was late in the game. They are still developing 802.11n products that deliver promises Ubiquiti made 4 years ago for the 802.11n rural crowd. For urban or suburban areas, the problem now is simply the 4-year thing.

Ahh, but Ubiquiti didn’t look back, they said, “Pshaw.” Forget about 802.11n GPS, what difference does it make (okay, my obligatory Hillary reference to failure). We don’t need no stinkin’ 802.11n GPS, we got us here an 802.11ac chip and we can make it communicate. We believe we can take that Atheros chip to great heights, including GPS. And if GPS doesn’t work, we can fall back on AirPrism which promises adjacent channel filtering that’s better than adding more CowBell. And after a rocky start with T-Tommy Wheeler changing the FCC rules to whatever he was told to do last year, Ubiquiti is finally getting DFS mostly working in 802.11ac. AirPrism is out and as for GPS, I’m not sure yet. Check your local listing for times and channels. I don’t plan on deploying it at this point so I don’t have any experience with the products.

Okay, so what does this have to do with Guerilla Wireless, Galactus – Destroyer of Wired Worlds, or whether HoHo’s are the same formula since restarting production (I think not and definitely follow the expiration date as gospel or else you will be wishing you ate garbage dump sludge instead of an expired HoHo, don’t ask me how I know). I’ll let you in on a little secret, Galactus is simply Guerilla Wireless all grown up. Guerilla Wireless was good, Galactus will be great.

Since cable went from 5Mbps to 300Mbps, pushing copper as far as it can go (I know it can go farther but the runs have to be shorter (dang that pesky, speed, distance, quality physics thing). The reality is the next step is fiber for the cable companies and some of the DSL companies if your area is worthy of it. That means there are enough people willing to spend $80-$100 per month for that privilege. If you aren’t worthy, and probably 50% of the country or more isn’t at our income levels, then you can get 12Mbps as the CenturyLink rep at my door tried to sell me, or you have to live with cable and its notorious, consistent as clockwork, rate hikes. At a minimum, a certain percentage of their customers are bailing on the bundle thing every year and watching streaming services: 1.7 million in 2013, 1.2 million in 204. It’s like the last guy watching NBC return the cable box to Cox or Time Warner.

Now add in the courts saying that Roku, Apple, and any streaming set top box can be a cable box, and well, you know that you don’t want to be caught in the middle of a battle between cable companies, content providers, Apple, Google, Amazon, and NetFlix. It would be like getting stuck between two Sumo wrestlers while dressed as a giant rice ball. I’ll take odds that the cable companies lose and you want front row seats if you are a WISP. And this is where all roads lead to: Galactus – Destroyer of Wired Worlds.

Guerilla Wireless was designed to bypass the big boys and offer a cheaper alternative. When the big boys kept arming up, poor Guerilla Wireless was hampered by a puny 98-pound weakling 802.11n chipset at the AP. And although the innovative dual-polarity omni antenna has had a good run, and we still use them, it’s evident that the future of doing battle with wireline providers is going to require not only 802.11ac, but 802.11ac with attitude, innovation, and the some heavy duty weaponry. The King is dead, long live the King means Guerilla Wireless, our Ubiquiti based 802.11n suburban based wireless network is being upgraded into Galactus, our Mimosa based network built around the A5-360 APs to start.
Don’t get me wrong, we have already started deploying Ubiquiti AC products in new areas. The combination of multiple long-range antennas and faster processors at the AP are an incremental improvement for sector-based designs. That’s not what Galactus is though and for that, the dual-polarity omni needed an upgrade, not only in concept but in innovation. That’s where the A5-360 comes in. Instead vertical or horizontal polarity, it used four directional circular polarity to improve multi-pathing in high-density environments and reduce noise. And with the C5 already using a slant-45 antenna, some quick field tests have shown not only a significant signal and s/n improvement over a standard omni, we are seeing it even surpass sectors that we have up in a NLOS environment.

So with all the announcements at WISPA America concerning Mimosa, Ubiquiti, and Cambium, where does that fit into our strategy and the industry as a whole? Keep in mind that even though I do rural where 10Mbps is a Godsend to some people, Galactus is all about bandwidth, baby. In that context, here is what we see happening in our network. And what I’m saying isn’t based on hypothetical scenarios; it’s based on the fact we have been running almost 30 clients (and still growing) for the last 2 months in parallel to our Ubiquiti 802.11n and 802.11ac networks. Basically, I think I’ve got a pretty good handle on how this is all going to shake out, at least for Galactus – Destroyer of Wired Worlds (I just never get tired of saying that).
This is the part where we get inside my head. Over the last 4 months, I’ve been trying to guess the release dates of hardware, firmware that is more stable than Astatine, (yeah, you know you are going to have to look that one up but trust me when I tell you that certain beta firmware we tested didn’t even run as long as it’s half-life), and still keep up with the tremendous growth we were experiencing. Now add in the effort to upgrade everything to support that growth without investing in something that is pretty much end-of-life literally in weeks to stay competitive. I can balance cups and saucers in each hand while riding a unicycle with a baseball bat balancing on my nose better than I could make these decisions, and I can’t ride a unicycle.

Soooooooo, in October, we started installing Rocket 5AC-Lites with the idea we can start transitioning our clients over from the Rocket 5Ms we had on the same tower. Yeah, that didn’t work out so great, someone forget to make sure the batter was stirred enough. So we backtracked and started adding more Rocket 5Ms APs to alleviate the load our existing infrastructure was getting hit with. At one point between October and January 1, we had as many as 68 users on one Rocket 5M and not too many behind that on a few others. In the meantime, I was watching the race between getting our Mimosa test site up and firmware from Ubiquiti for the Rocket AC5-Lite radios to talk to 802.11n radios. Simultaneously, we were having some random problems (never solved) in another part of the state where we deployed all Ubiquiti AC products. I saw the problem go away with 7.2 firmware so that pretty much told us what the issue was. The Pepto-Bismol was flowing like the Mississippi river around our office.

My Christmas present came, not from Santa, but from Mimosa when a bunch of C5s and A5s showed up on my doorstep. We immediately started offloading some of our users to the A5s by swapping out their Ubiquiti radios in one area with C5s. Even though they were early beta, they worked right away. The Rocket 5Ms those users were attached to, was handling almost 50 users, so it was straining hard in the evening. After transferring about half of them, we had that particular Rocket 5M sector down to about 26 users and the same amount on the A5. Let the testing begin.

Yeah, I know that testing an 802.11n radio against an 802.11ac radio isn’t fair. Unfortunately, you can only test what you got. We did do a quick test between a Rocket 5AC-Lite with a 17dBi sector through trees to a Powerbeam 400-AC and an A5-360-14 to a C5 with the its internal 20dBi antenna. We didn’t have the backend capacity at this location to do more than 120Mbps and we also had the issue with firmware on the Ubiquiti causing some weird anomalies, so there was no way to do any side-by-side testing other than straight signal levels. And keep in mind this was early released firmware for Ubiquiti with only 5.8Ghz available and the Mimosa was on Day 1 beta firmware release. So given those limitations, we did check the signal levels between the radios and the PHY layers and we noticed this, through the trees and compensating for rated antenna gain numbers and setting the power to fixed levels on both radios, there was a was signal improvement of several dBi for the Mimosa pair. Without more testing, we can only guess one of a few things, antenna ratings aren’t matching the reality, power outputs aren’t really accurate, the combination of a circular polarity antenna connected to a slant-45 antenna works better either LOS or through vegetation, or the noise level was slightly lower for slant-45 but still leaves us with 3-6dBi difference we couldn’t account for. I can come up with a few more ideas but further testing in another location comparing a 19dBi sector with the Rocket 5M to a Nanostation 5M-Loco showed as much as a 10dBi+ difference with even more trees. In fact, the picture of this shot is on the front of our web page.

There are a lot of variables that could cause these differences so please, save your emails. I’m an engineer, I tried to minimize them the best I could to make the tests as close as possible within the limitations of time, effort, and frankly, my desire to get this beta test up as quickly as possible. What we did do, and again, this is the A5-360 14 against a Rocket 5M with a 19dBi AM-120 sector on the same pole, was to check signal levels as change out M radios on the roofs. In every case, the signal got much better and our results speak for themselves. Speed tests have no value here since we were also testing 80MHz channels and keeping a live system operational. We posted those results on Facebook and LinkedIn.
So you ask, where do you go from here? Well, all I can tell you is that it depends. We have been testing the A5 in areas where we replace short-range M radios and new areas with longer ranges of up to 1.5 miles (The A5-360 14 isn’t really designed for this, that’s where the A5-360 18 comes in but just for testing purposes, it hit -64dBm and as high as 360Mbps (PHY Layer). We are also testing Ubiquiti M radios talking to the A5-360 while simultaneously testing Rocket 5AC-Lites in the same scenario (hint, the firmware not to be mentioned is stable as of last Friday night, latest versions on the M and the AC side, and totally usable as we just converted 25 people over and it’s working well). So now with Ubiquiti having a stable migration path from 802.11n M series to 802.11ac, there are a lot of options.

In our case, where we only have to keep delivering up to 20Mbps, we are staying with the M radios and upgrading the APs from Rocket 5Ms to Rocket 5AC-Lites. In the areas where we have to deliver up to 50Mbps, we had already started deploying Ubiquiti AC radios and will stay with those for the time being. The latest firmware, 7.2 seems to be working very well for AC areas only. There is no upgrade path from Ubiquiti AC radios to Mimosa as AirMax is built into the chipset and can’t be turned off like it can in the M series. In new areas where we are head-to-head with cable and DSL providers and our service offerings start at 75Mbps, we are moving to Mimosa. Since we are pushing services up to 200Mbps and we have tested as high as 340Mbps in a real-world environment which isn’t even close to the full capacity of the AP, this will be our future against the wireline providers.

Just to add a wrinkle in all this, keep in mind that our entire Galactus model is based on the customer ROI being in 2-3 months, not 18 months or ungodly fiber ROIs of years. It can be deployed without millions of dollars of subsidies the government so freely gives to CenturyLInk to overrun underserved areas. It’s also based on a strategy that can deal with low to medium vegetative areas as our NLOS testing of the Mimosa gear has shown so far. Unfortunately, until the politicians (I wanted to use a harsher word here, but my wife, the censor, stopped me) decide it’s cheaper and more productive to give unlicensed spectrum to the free-market and let innovation and American ingenuity find better ways to deliver services than give it to unproductive companies like CenturyLink, deployments like Galactus — Destroyer of Wired Worlds — will be the exception, not the rule. However, just to be a complete jerk about the whole thing, we upgraded areas where CenturyLink got $500,000 to put in a whopping 25Mbps service with that still isn’t installed, with ahem (clearing my throat), 200Mbps IN YOUR FACE SERVICE starting at $55 for 75Mbps and going up from there (we have cheaper options coming when we can get more equipment). Coincidentally, the ROI on the customer side is 3 months depending on the mount. Wow, no taxpayer subsidies, what a concept. Some bureaucrat’s head must be exploding right now.

I am not sure what more you need to get started. The tools are there, Mimosa is real and has laid out the future, Ubiquiti is warming up more guys in the bullpen, Cambium is filling in NLOS holes with new equipment based on current product lines, and there are more backhaul options than I have baseballs in my garage. It’s time to quit watching NetFlix and start delivering it.