Tales From the Towers: Chapter 61 – Migration is for the Birds

We’ve talked about migrating from 802.11n to 802.11ac or from one manufacturer to another and all the pains involved with that. Sometimes it goes as smoothly as putting on a bandaid, other times more like ripping it off, taking the scab with it, and then it keeps throbbing for days after. Before I make you waste a half hour to see what we eventually decided on, I’ll summarize it here. If I had written this a year ago, it would have had to be censored for all the cuss words I would have used. Loss of sleep, ulcers, and paranoia that makes you think the APs are alive and plotting to get you pretty much sums it up. If you are starting your migration today, then you will firmly believe that this is a piece of cake. It might be a cake that costs $30 per slice if you are migrating to Cambium from Ubiquiti, but it will still go down easily with a glass of milk. But like my mom always said, either quit complaining about the broccoli or no cake. So let’s break this down within manufacturers to start with.

If you are migrating your Ubiquiti or Cambium equipment from 802.11n to 802.11ac, it’s as easy as pie (starting to catch the theme that I’m hungry and want desert yet?). With Ubiquiti, just drop in any 802.11ac Rocket to take over from your tired and overworked Rocket 5M and all things being equal, speeds to even your Nanostation M5 Locos will go berserk. In fact, the one thing you will notice is that your M radios can really deliver 60+Mbps to residential users with a 20MHz channel when they aren’t being choked at the AP. Then you wonder why this wasn’t addressed years ago. Trust me, it was mentioned on more than one occasion with multiple vendors.

This particular bottleneck had been a huge sore spot for me and one of the reasons I went to Mimosa Networks for part of our network. Mimosa Networks promised cross vendor compatibility (and even the ability to talk to 20/40/and 80 MHz client radios simultaneously), meaning faster APs before Ubiquiti got the firmware worked out for 802.11N compatibility. This worked but had some caveats on range and that because of that timing, the Ubiquiti client radios needed to be tweaked individually. In the meantime, Ubiquiti was still working through the firmware development of 802.11ac backward compatibility and complete functionality while I got a couple years older and my little orange tree finally gave up some fruit for the first time since I planted it (look up how long it takes for a small orange tree to bear fruit on this one).

Cambium watched from the sidelines and thought, never let a crisis go to waste. Especially if you can squeeze a few bucks on the licensing and figure out how keep the GPS protocol and work with Ubiquiti clients. I love capitalism and the options it brings to the market and if you plan on going Cambium, it was a great move. If 802.11ac doesn’t interest you yet and you want to start a Ubiquiti deployment to a Cambium migration, then Cambium’s licensing program so you can run their ePMP APs with their GPS sync and faster processors was brilliant. When they eventually get their 802.11ac stuff out, I’m sure it will be backward compatible as it would be insane not to. At the same time, they are probably struggling with it as much as Ubiquiti did over the last 2 years, especially if they have to make it compatible with 802.11n Ubiquiti equipment. I’ve come to the conclusion that the phrase, there has been a delay in the firmware, is probably taught to every Engineer in their freshman, oops, Freshpersons year (I’m nothing if not politely correct).

One of the problems with WiFi compatibility across vendors is simply that WiFi mode is not designed for longer range due to timing issues. When we deployed the Mimosa Networks A5-360 14 with Ubiquiti clients to upgrade Rocket 5M’s with sectors and RF shields, we had to deal with 2 issues. The first was simply noise. The A5-360 14/18 (we tested both) was picking up a lot more noise since it has a 360 degree coverage zone. Users that had weak signals were hit hardest by the noise. Users that were more than about 1.5 miles also had lower modulation rates or were varying modulation badly. We eventually had to turn off Auto-ACK on the Ubiquiti CPE’s and set distances manually which fixed most of the users. We are still deploying A5s and A5cs to upgrade parts of our network which are moving to full Mimosa through attrition, but the criteria is users are less than a mile away, and if there are noise issues, A5cs with RF Element antennas. We have Mimosa A5’s with more than 60 users on them with no issues under this criteria in WiFi compatibility mode and even my personal favorite, the little CPE that could, the Nanostation 5M Locos are pulling 50Mbps. Keep in mind that the distance issue doesn’t apply when you are using Mimosa C5 client radios, regardless of compatibility or TDMA modes on the A5s.

In some cases, we kept the Ubiquiti APs on the pole and put the A5-360 or A5c above it since we wanted to move onto TDMA testing on the Mimosa. New clients went on the Mimosa, Ubiquiti clients stayed on the Rocket 5Ms which are at minimum getting upgraded to Rocket 5AC-Lites until they are moved. Fortunately, the C5c radios are now shipping so that extends the range of the A5-360 on the s/n issue. Still working out the antenna choices though but there are several out there from Ubiquiti, RF Elements (another soon story), IgniteNet, KP Performance, and a couple other companies that I’m not endorsing yet until I do further testing. All of these are usually pretty good performance and the mechanicals are solid.

On the other hand, LanBowan for example, looked like a good candidate and met my goal of cost, weight, and reduced wind load since it was a grid. Quick testing on the pole showed the gain was a couple dB down from advertised which we realistically expected based on size. Many manufacturers like to push those numbers using various non-standard marketing, testing that is umm, let’s say optimistic, or possibly they just pull numbers out of a hat and that’s the one they go with (Yes, you know who you are). My favorite is the 15dBi Rubber Duckie antennas we use on the AirRouter HPs (It’s probably closer to about 4dBi but definitely better than the one that comes with it by about 2-3dBi and $2.50 in quantity. What’s funny is even those antennas have to be tested because one Rubber-Duckie we got was actually worse than the included antenna). We will be testing the KP Performance and IgniteNet 5GHz antennas shortly also when they get here.

Migrating to 802.11ac APs at minimum is pretty much a no brainer those of us battling wireline or WISP competitors. Rockets 5Ms, although reliable, are past their prime. The next logical options are Ubiquiti Rocket AC and their Prism products or Mimosa. Cambium and their ePMP products are still 802.11N but for remote areas, that may be more than enough. For those of on the march to Infinity and beyond, I mean 200Mbps and beyond, 802.11n is like hanging out with our grandparents.

One of our competitors just deployed new ePMP 2000 equipment and Medusa on a water tank in an area they just took over for example. In the same area though, we are deploying Mimosa gear with TDMA and 80MHz channels all around it, kind of like surrounding the castle and playing AC/DC 24 hours a day. The noise battle is going to be intense but physics, geometry, and a carefully monitored Capex are going to make this an interesting test case. Even I’m excited to see this one play out. Anyone want to take bets on the results?

We are overlaying areas where we know our competitors are using last generation equipment with new equipment and taking away customers from them because of it. Our deployment model is also changed to future proof our network. This makes it prohibitively expensive for others to catch up while also hopefully protecting it against future cellular interference (think pie again, and you just thought I brought up desert because I’ve got leftover Banana Cream Pie in the refrigerator). At the same time, we make sure that the ROI of any area is 8 months or less and each residential installation is 91 days (Education Everywhere is the exception, slightly different financial model) or less so we aren’t roadkill if the next great technology comes in to compete. If we can’t meet that model, we either change the model technically to meet it or we don’t do it.

For those of you in rural areas with little or no competition and ARPU’s that are insane like $90 per month for residential, this really doesn’t apply. That is unless Verizon or AT&T moves into your area without data caps. I know a lot of you have seen this and lost or gotten customers back because of it, but the cellular companies are sneaky. If they have a tower in rural areas that isn’t being hammered, they seem to be selectively offering unlimited home service to those people. I ran into this after knocking on doors a few months ago (I think it’s good to go out to assess new areas), so ended up modifying the pricing a little.

Getting back to the migration issue between vendors, let me sum it up this way. If you are happy with Ubiquiti or Cambium and plan on migrating to 802.11ac, the migration problems over the last 2 years are pretty much over. With Ubiquiti or Mimosa AC, it isn’t going to matter how you deploy, it just works. Stay with them and have a long and prosperous life. You will have to wait a while longer for Cambium to get their 802.11ac products out of the oven If you aren’t happy, then keep the old equipment running and simply add your new vendor and run them simultaneously until you cutover all the clients. There are unique situations where you don’t have to do that, but in general, that’s as good a plan as hiding the leftover pie in the garage refrigerator after holiday which your wife doesn’t go into very often. I just say I took it to the office….

On a quick side note, we finally have a 15 mile shot set up to definitely test the Mimosa B5c versus Ubiquiti AF5x versus Ubiquiti Rocket Prism versus Rocket AC-Lite versus Mimosa C5c PTP. No B.S., same channel for all of them and then split channels for the Mimosa B5c and the AF5x. No vendors involved, we have all of the equipment in stock. I’d add Cambium but I don’t have it in stock but if someone wants to donate a couple off the shelf for the test, I’ll throw it in there. We are running a pair of Powerbeam 620AC’s on that link to start the testing and they are pulling 42Mbps average on the connection (haven’t done TCP/UDP tests but we will with Mikrotik routers at both ends) with either 10 or 20MHz channels (gotta love interference). With the latest firmware, the Powerbeam 620ac’s are a great value for PTP so it will be interesting to compare them to everything else. All the other radios will be doing RD-34 antennas on one side and an RD-30 or whatever the equivalent gain model is today on the other side (having a hard time remembering all the new antenna models and their names these days there are so many). The interference means that test won’t be entirely empirical since we can’t control that but it seems to be pretty consistent so it will have to do.

Word to the vendors: please don’t start having your lawyers send me letters not to publish my results as this will get published along with the results, ALL OF THE RESULTS!!! Freedom of the Press and all that and my attorney says it’s okay. If you want to send techs to monitor the test, fine. I’m pretty confident that we will do as fair a test as possible or I wouldn’t be discussing this publicly. There will be no beta firmware, no special sauce, same LMR-240, same length, etc… If there are anomalies or we have to retest a radio because something looks like it’s changing, we will do that. We will definitely check all radios twice at different times to try and get an average to try and compensate for changing interference possibilities. Antennas will be pre-aligned and no changes once the test starts, etc… However, if your monitoring techs bring diet coke, in a cooler full of ice, well, we might let you stand closer to a laptop for a better view. It’s going to be 110-120 degrees when we do this and you will have to climb up 160’ of stairs on a grain elevator so don’t send anyone who can’t survive this. Making us call the ambulance to delay the test is not a good tactic. Let the best radio win….

About Rory Conaway

Rory Conaway has been in the IT and Wireless Industries for the past 25 years as an author and consultant. He currently operates a growing WISP operation in Southern Arizona. He consults with investors, manufacturers, and WISPs, and develops financial business models for startups. In addition to writing articles in industry publications such as Mission Critical Magazine, Mr. Conaway also writes the series “Tales from the Towers” that can be found on various such as www.triadwireless.net and www.muniwireless.com. He has also engineered several wireless designs such as S.P.I.R.I.T. and Guerilla Wireless as well as building integrated wireless and video surveillance for airport security, municipal and critical infrastructure, SCADA systems, and hotel/MDU deployments.


  1. Mike Hammett says:

    You have grain elevators in Arizona?

  2. Nice article Rory. I would put my money on ePMP and Medusa vs. 80Mhz channel Mimosa.
    Let us know the results.

  3. Rory Conaway says:

    LOL, yea, I’m sure Mike. With C5c’s and some other secret sauce still coming out, you might want to save your money. First though, comes da da da da, (drum roll), “THE TEST”. The first antenna is going up today.


  4. Great article–always enjoy reading your posts!

  5. Jens Jönsson says:

    80 MHz channel width. Where on it is that possible ? Install a neighboring AP and it’s killed…

  6. Rory Conaway says:

    Everywhere for the models we are deploying this in. Speed, distance, quality, pick 2.

  7. Nice One Rory ,,

Speak Your Mind