I need to wrap up the Cambium and Ubiquiti comparison so this is the end, at least for a while. Part of the reason is that a third player is now in the game, Mimosa, so things are changing. I know there are other companies out there like Bitlomat, Proxim, and Radwin, but they are a very small part of the WISP market and I doubt that is going to change. They are all also proprietary and I’m not going in that direction with my philosophy although Proxim is working hard on integrating other GPS systems. Not sure about the other vendors.
Chapter 49 got just too long to cover the nuances of all the announcements from Ubiquiti and Cambium so filling in a few holes might be in order. Migration, industry compatibility, new technologies, and how to integrate those technologies into new business decisions needed to be explored with all the announcements at WISPA. Each one of these areas could take a thesis to cover but since my wife gave me a Honey-Do list as long as an Obama speech without a teleprompter, I’ll try to keep mine short.
I’ll start with what was bugging me after re-reading Chapter 49 (Ubiquiti versus Cambium, Round 2). I thought clarification was seriously needed on my thought process about preparing for 802.11ac. One of the main problems for Cambium users who started with Motorola Canopy was an upgrade path. From the really clever PMP100 FSK product line, Motorola went to the PMP430 product line and then jumped on the WiMax bandwidth with the PMP320 product. No compatibility with anything previously released. When the PMP450 was announced, Canopy users were told that it was going to be compatible. Then they found out it wasn’t going to be compatible before it was compatible but then it became compatible, at least with the 430 and by sticking a second head on the shoulder of the 430, compatible with 100 series. Although all of those operators stayed Cambium, many operators went looking for additional options, a la Ubiquiti. The PMP450 also ran into the pesky physics issues of OFDM not going as far as FSK since it was about 6 times faster. The old adage, speed, distance, quality, pick two, still makes a good motto to fly by if you are changing modulation schemes. 802.11ac and/or 256QAM won’t be any different.
Cambium operators who tried Ubiquiti not only ran into the same thing, they also found out that GPS on Ubiquiti was kind of like a blind date your friend set you up with. The reality just didn’t match the hype and the results were as close as Mondale’s campaign against Ronald Reagan. It just didn’t really work very well or much at all, depending on who you ask. Cambium users used to a functional GPS were highly disappointed, as was pretty much anyone who never even used GPS and expected results based on Robert Pera’s presentation discussing it.
The ePMP product line was a coup for Cambium to get it out to market since it was their first foray into a standard chipset. As good engineers are wont to do, sometimes at the cost of market share, they fixed one of the problems with Atheros and 802.11 that Ubiquiti still hasn’t figured out – GPS. Unfortunately, as I’ve mentioned before, that made it incompatible with anything else. Recently, they just announced that the ePMP CPE units will have 802.11, which is huge. How huge is it? Glad you asked. It’s so huge, that I had to delete 2 paragraphs and rewrite this one during football Sunday which didn’t make me happy (and there were the ribs I was planning to cook). Cambium makes a high quality, well-engineered product that has unique features like heaters in case you live in any area that I will never move to for that reason (Phoenix is so SUNNY!!!). My thesis, which I promised not to write, was based on the fact that if Cambium took a few years to come out with 802.11ac, you would be kind of stuck.
With this announcement, Cambium is finally assuring their customers that they don’t have to forklift an entire network to move to future technologies if Cambium moves at Motorola speeds. With 802.11ac knocking on the door (based on FCC filings, Ubiquiti’s Rocket 5M AC, WISPAPALOOZA announcements by Mimosa and the full listing of their product line on their website, www.mimosa.co) , and past experience with Ubiquiti firmware, 802.11ac is not going to make an impact for several more months in the PTMP market and that’s a comforting thing to know. And with Ubiquiti, you know they are going to be 802.11 compatible since they married Atheros a few years ago and are now having an affair with Broadcom for 802.11ac. For rural operators, it won’t make a difference. For the model I put forth in Chapter 43 and future articles where we try to keep clients at less than 1 mile from an AP, 256QAM isn’t an option, it’s a matter of survival.
Hopefully Cambium has something up their sleeve or what happened to Motorola when Ubiquiti came out is going to happen to them. Ubiquiti created a whole new market at the low end that Motorola never tapped that resulted in massive growth in the industry. Now that they have something to compete with, albeit 2-3 years too late, they not only have to compete with Ubiquiti’s new products and what they are already shipping, but Mimosa announced features nobody else even imagined in next generation products. What’s worse for Cambium (and Ubiquiti) is that those features are so out of the box, they were never even discussed in house, let alone a possible product. And the mechanics of the Mimosa omnidirectional APs which are at minimum 5dBi better than anything else from any vendor, rely on a chipset that is already ahead of what Broadcom or Atheros are shipping, so duplicating it isn’t going to happen easily. Even if they get started tomorrow, it will take them another 2-3 years to get product to market based on their track record. When Cambium split off, I thought it was going to be a new kind of company, but when the ePMP came out and the parabolic dish a year later (seriously), it was evident that the old engineering mindset never changed. It’s better than Motorola, but Mimosa just dropped a dime on them and they are going to have trouble picking up.
In the case of Triad Wireless and our new sister company, Cat Mountain Wireless, the sole reason for staying with 802.11 compatibility is so that we didn’t have to change the entire network to take advantage of 802.11ac or 256QAM. Also, based on what the world now knows, Mimosa is showing products that can support up to 1Gbps on an AP along with MU-MIMO late next year (with a 4×4 system, MU-MIMO will at least double the throughput in a PTMP system by talking to different 2×2 clients on different streams). Our system can live without GPS or polling as the network was designed around that idea, although polling has some attraction. I believe it’s more important to make sure that I have an upgrade path with multiple vendors. Ubiquiti has also already released an 802.11ac Rocket that works in PTP mode (now I’m interested to see who gets their PTMP AP working well first, Ubiquiti or Mimosa).
After having Vivato, SkyPilot, and Ricochet leave me at the prom, I’m not taking any chances. Most WISPs try to get more than one supplier for their bandwidth but don’t’ consider multiple hardware vendors an issue. Just to make sure to keep this in the back of your mind, Alvarion filed for bankruptcy protection last year. And although my paranoia is limited to the efforts by Major League owners to prevent the Cubs from ever winning the Pennant (they still say the person who interfered with the fly ball during the playoffs was just an overzealous fan who made a mistake. Yeah, I believe that!), I’m not taking any chances. Another story to keep in mind is that when Chevron-Texaco bought the rights to the Nickel-Metal Hydride Oshvinsky battery, they pretty much killed the electric car industry for 8 years and then had the gall to sue Toyota for using it. This one really is true. This is a competitive industry with let’s just say, some aggressive attitudes that have been made rather public and billions of dollars at stake. I suggest having a backup plan for any single supplier is probably a good idea.
Designing a network that can compete against wireline is a challenge. Vivint is pouring boatloads of money using APs that Huawei is helping develop (probably not Vivint’s best idea from a public relations standpoint since Huawei has been banned from the cell phone industry in the United States due to their funding and ties with the Chinese military). Their model is based on building their own proprietary APs to deliver 50Mbps for $50. They use houses as relay points which I think is a pretty good idea since we’ve been doing it for about 7 years. It’s really the best financial and logistical option in many suburbs for wireline competitive speeds unless your brother-in-law is the City Manager and your last name is Daley.
Vivint is guessing though, that for every rooftop they get, they can connect a certain number of indoor clients. Vivint doesn’t want every customer to require a truck roll which is a good theoretical strategy but not very practical from a financial or technical position. My guess is that they thought they could get 10-15 clients to 1 roof mount antenna based on the performance of the Quantenna chipset. I’m also thinking that in reality they are lucky to get 5 to 1 in some areas or worse. Our network has a ratio of 4 houses for every fixed commercial asset (towers, buildings, etc. which feed the houses with backhaul) for example and although we are up to about 30 users per house/AP, that’s just all we have sold to so far. Currently the model could easily support up to 50 clients per house with a single Rocket AP. However, aesthetics and home-owners associations come into play here so multiple APs with sectors (and totally forget shields here) aren’t a viable option. This is where I think the new Mimosa AS-360 products will have their best application. Theoretically, not only will it handle a plethora (Three Amigos – hilarious) of users and capacity, it won’t look like I have to get NASA clearance to install.
To that end and to expand the foundation for taking out wireline providers, we have started a full upgrade of our core network to improve speeds to be more competitive with wireline at a better price. For example, we had 10MHz channels on our 5GHz APS and have changed them to 20MHz. At the same time, the client APs are being set to 20/40MHz for future expansion when stable and tested 802.11ac APs are available. The original reasons for 10Mhz channels were:
1. We don’t run more than 50 users per AP since we have lots of APs (with the Barracuda Web Filter and blocking of certain applications, we never saw the problem at 30-35 users others reported).
2. Rocket 5M’s never got a functional GPS so channel spacing was an issue. The general rule is double the channel width as a guard if you don’t have GPS or get RF Armor shields.
3. The Ubiquiti processors on the Rocket 5M were limited to about 22-25Kbps and less with AirMax. This could be almost achieved with 10MHz channels and a lot of small packets.
4. We only needed to deliver about 35Mbps per AP which was sufficient up until last year.
5. Smaller channels, less guard space, and we were still dealing with a lot of Legacy Pre-N Ubiquiti and other 802.11 equipment that didn’t support DFS which limited the bands.
When the upgrade is completed over the next couple of days, users should see peak speeds of 50Mbps which isn’t bad for 802.11n. Some of my customers are already reporting a doubling of speeds to 40Mbps from 15Mbps and we haven’t finished all the backhauls yet or upgraded the tower.
Part of this upgrade is replacing the PowerBridges with PowerBeams set to 40MHz. Although I’m kind of ticked that the PowerBridge never got DFS or the 5150-5250MHz band, I’m replacing them with PowerBeam 400’s anyway for the speed. This is very sad since I’ve never had a single PowerBridge ever fail all the way back to the original PowerBridge 2’s. The PowerBeam 400s are so much better and faster for less than half the price; it’s time to retire the PowerBridges. Although the Rocket ACs are sort of out, waiting for some future firmware upgrades is probably not a bad idea. It seems that manufacturers today are releasing beta firmware with hardware more often and being an early adopter is probably better left to the lab than the field. Timing hardware release and stable software release seems to be harder than the Cubs winning the Pennant. The battle with cable/DSL/NetFlix is never-ending and in anticipation of a future with 802.11ac this is a great time to prepare if that’s the direction of your network.
A few side notes:
I’m pretty impressed by some of Ubiquiti’s new equipment. I really like the innovative design behind the AirGateway- LR’s. Although they need a sensitivity level setting like the AirRouters in the firmware, they not only added to our bottom line, they have so far shown a far lower failure rate than the AirRouters. Our profitability per install went up $6-$18 (we offer an optional longer range antenna) each with these little gems. I think we have about 200 or so AirGateways deployed now. $20-$30 for a managed AP opens up a lot of ideas although the standard AirGateway without the antenna has the range of a strand of spaghetti. In some homes though, we have had to put 2 of them in to cover all of it, even with bigger antennas. Bartender, AirGateway-LRs for all my men!
The new NanoBeams have been almost perfect (perfection comes with DFS frequencies which aren’t there yet). In PTP modes, especially shots through trees, we are not only seeing big improvements in the connection quality over PowerBridges and NanoBridges, the faster processors are testing out well over 100Mbps as we expand channel sizes. They aren’t hitting the full speed of 802.11N with 40MHz channels, but it’s much closer now to the ePMP. 802.11ac though, will be a whole new ballgame. Expect at least 250Mbps of real world throughput with 40MHz channels. Along those same lines, the new NanoBeams also look great. The mount options may save $10-$25 per house in some areas and labor time while the faster processors mean at least 80Mbps or more.
For you Cambium fans, Cambium has finally released a CPE that doesn’t look like a SETI experiment for longer range. Can you tell I’m not a big fan of reflector dishes? It’s an old tired design that adds nothing to the bottom line that an integrated parabolic could do better and it needs bigger mounts and more bolts into the roof. Cambium finally agreed with me after 3 years and built the ePMP100 at a cost of only $15 more than their standard dish. Then they topped it off with a PTP ePMP Force 110 which uses GPS synched radios as a higher-end competitor to a pair of PowerBeams. It fits in price wise between the PowerBeams and the AF5 or the Mimosa B5 Integrated. I might have to do a speed test someday and see how they compare.
For the fastest unlicensed backhaul though, the real battle is about to begin between the AF5 and the B5 Integrated (Cambium has a new 2Gbps 820 radio also). After Robert Pera threw down the gauntlet and said this,
“So Mimosa’s radio is coming in at a price point similar to airFiber5, but it’s really a hack-together Wi-Fi radio. And it’s not even 4×4, it’s 2 bonded 2×2 radios each 80 megahertz channel, size of channel. And then you need 200 megahertz of continuous spectrum to reach those speeds. And not only that, but only one of the chains is doing DSS radar detection, so it’s impossible for them to be compliant to anywhere in the world and have that contiguous 200-megahertz spectrum channel. So I see them pretty much as an alternative to our AirMax AC point-to-point solution, but the AirMax AC point-to-point solution is essentially 1/4 the cost. So again, I think they’re very good at branding, they’re very good at hype. But at the end of the day, they put lipstick on a Wi-Fi tape.”
The “hack-together” comment was more like a glove to the face so I will be testing those radios when I can get my hands on a B5 to find out the truth. I don’t see them as competitors though, I see them as complimentary. The Rocket AC has more speed than a Titanium but I don’t see it handling noise as well as the B5. It also has to deal with the inherent differences between 64QAM and 256QAM. The AF5 has more throughput than either radio with its 1024QAM modulation scheme and FDD but the same problem in a high noise environment and tears up most of the 5.730-5.850 band if you want long range and 1Gbps. The B5 will handle noise with different techniques so the tradeoff here of speed versus a high-noise environment will be interesting.
FTTH or at least 300Mbps cable is coming or is here. There is no way around it. If a politician is willing to spend $14,000,000 per customer for Obamacare, $10K to the home probably sounds like a bargain to them and it pays off their donors. It doesn’t matter that the return on investment will take longer than the house will stand but we have politicians who still think that’s a good “investment” of taxpayer’s money. I’m sure they could find a whole lot of people in Detroit and other economically depressed areas who really want to pay $100 per month for these high-speed circuits. Apparently the need to run vaporware is endemic in the psyche (my tongue is so far planted in my cheek on this one, I’m using it as neck scratcher). I will bet anyone that 50% of the market or more would be happy to pay lower prices as long as they can watch NetFlix without buffering in HD, regardless of advertised speeds. However, most users are going to eventually expect massive capacity for reasons they don’t have a clue about. If their neighbors have it, they will want it. Not everybody of course as we are targeting the 50% who just want good service for the best price. We are also noticing that although bragging about speeds is a great advertising gimmick, bragging to your neighbor that you are saving $100 a month is an even better sales tool.