From Integrated Product to Challenging Goliath
How airMAX AC Generation 2 Technology Completes our Industry Vision
Ten years ago, Ubiquiti was just a new entrant in the WISP Industry, known to be little more than a manufacture of niche mini-PCI radio cards that could be used with Mikrotik router boards. At that time, I saw the writing on the wall as Mikrotik was starting to bring to market their own radio cards. They had a complete monopoly on the growing WISP Industry with a decade head start and an operator customer base they built around their routing software foundation. However, I saw a critical flaw in their strategy that could possibly be exploited. Mikrotik believed in an after-market assembly world where they would market hardware and 3rd party companies would supply the mechanicals, antennas, and cable components for WISP’s to assemble solutions themselves for deployment. But, I saw the future differently — I believed WISP’s would welcome product integration with a greater attention to user experience.
Ubiquiti at that time was made up of just a handful of people. Myself and Patrick Jabbaz worked together on the schematic design, he owned the layout and digital design while I owned the RF Design, radio tuning, and manufacturing testing. Also around that time, we put together a small talented developer team in Lithuania who worked on our firmware called “AirOS” and we hired a mechanical design consultant as well as an antenna design consultant which would help us in our first foray into manufacturing a complete product. Like all airMAX products since, I have taken an “outside in” design approach where I worked with the designer on my vision for the product first and then we customized the hardware and antenna to accommodate.
This project we would market as “PowerStation” and I hoped it would become an Industry standard for CPE deployments. The initial response in our U.S. channels was positive and we received orders for several thousand units. It was at this time, I took my first trip around the world with a PowerStation sample in hand to try and garner international channel interest. Within a couple weeks, I would travel from California to South America and Africa, then on to Russia, the Middle East, Europe, and finally on to Asia. By the time I was finishing the trip, I had become quite demoralized. While the PowerStation design was acknowledged as being beautiful in appearance, it was universally criticized as very expensive, bulky, and impractical for the various emerging markets we were targeting. Why? The Power station was large and heavy making it problematic for shipping and with a manufacturing cost of nearly $100, it just could not be competitive in these markets where solutions assembled using components from Asia were being sold at nearly 1/4 the price.
Even worse, after I got home and shortly after shipping PowerStations into the U.S., we began seeing numerous quality and performance issues from firmware crashes to reports of our high-gain antenna not having any gain at all. The situation was a mess — I was frustrated having little gained from all the development efforts, nervous at the prospect of not being able to compete in the market as Mikrotik was quickly moving to lower cost solutions and solidifying their lead, and concerned financially — speculating how much this misstep would cost us as Ubiquiti was a completely bootstrapped operation.
But sometimes your best ideas and achievements arise in times of great adversity and in those next days, we took a hard look at the PowerStation design as well as what I learned from my recent travels. What struck me now about the Powerstation’s mechanical design (which followed traditional practice) was it’s over-complexity with several plastic parts, various waterproofing gaskets, and heavy die cast parts for the antenna ground plane and hardware enclosure. Coincidentally at around the same time, Motorola had a popular product gaining attention called “Canopy” that was priced far too expensively for the emerging markets that I sought after, but was very successful with higher-end WISP in the U.S. What intrigued me most about their design was a very simple plastic bucket enclosure that was light-weight, low-part count, and eliminated all of the complex weatherproof gasketing.
Within the next 6 months, Patrick and I executed on a new product idea from concept to volume shipping (an outstanding achievement) that would be known as “NanoStation” Where as PowerStation was bulky and over-engineered, NanoStation exemplified minimalist design. We found new mechanical engineering resources and inspired by Motorola’s direction, made the industrial design even sleeker as well as took advantage of the full PCBA backside and using it as a ground plane for a cost-effective dual-polarity directive antenna array (which was designed successfully this time). Plus we even integrated a software selectable external antenna connector. We essentially added the directional antenna and weatherproofing around the hardware for only a few dollars in a way that integrated seamlessly. We also had the fortunate timing of solving critical SW issues on the heels of the failed PowerStation experience.
As soon as we had NanoStation prototypes ready, I went back around the world to pitch the new product to the same international channels, and the demand was overwhelming. Supply constrained, we would ship over one million NanoStations within that next year — it was the company’s first growth inflection point.
airMAX Software Strategy Coming Together At Long Last: A Complete Set of Software Technologies
Bringing NanoStation to the WISP Industry put us on the map, but our ambitions have always been much greater. We have never settled for being a typical equipment provider. We do not market to big telecom carriers nor have sales teams. Instead we opted to disrupt the status quo. We wanted to give the tools to any ambitious entrepreneur — even ones without financial backing or technical training — to compete in the big carrier dominated ISP Industry. We created a community platform where these very entrepreneurs could share experiences to help make our collective technology better. One way I like to think of our strategy is essentially supplying the slingshot in David’s fight against Goliath.
Unfortunately, as many of you know first-hand, we have had bumps along the way towards reaching our end vision. But just like how we hard pivoted successfully from our failed PowerStation product to NanoStation, we have made hard pivots in our development strategy over the past couple years that have achieved outstanding results.
On the software side, I believe we have finally achieved a true end-to-end solution for operators to not only deploy and manage their networks, but also to scale their business. This impressive collection of software resources Ubiquiti provides free of charge in an effort to “level the playing field” against larger carriers with more financial resources.
From link planning (AirLink), to deployment ease (U-Mobile), radio optimization (AirOS), network management (AirControl), and business operations (UCRM), we are empowering this Industry with an unprecedented software technology arsenal.
Radio Configuration and Link Viewing — The New AirOS
Nearly all WISP products (including prior airMAX/AirOS technology) have UI design roots based in the WiFi router world. But where as WiFi routers are deployed in indoor environments, outdoor wireless is about long distances links focused between two points in a widely varying and difficult to assess outdoor environment.
Looking at things from this perspective, I realized a good UI for WISP applications should sharply focus on providing insight into the outdoor environment and how it affects a specific long distance link.
Our development team has done a fantastic job in creating by far the most impressive WISP product UI brought to market. The new AirOS quickly gives insight into link distance and settings, capacity in each direction, and air-time. In addition, it gives the operator an instant overview of the total spectrum behavior (powered by our dedicated AirView “always-on” spectral analyzer radio) as well as provides fantastic insight into the relationship between signal strength, modulation, and capacity. Finally, cumulative distribution charts give insight into the links time-dependent variation.
Radio Deployment — The New U-Mobile (both Android and iPhone support)
In less than a year, UniFi mobile has in excess of 120,000 active users and over 1 million downloads! So, the thought occurred to us — let’s put the same team on building an equivalent app for the airMAX world! With this plan in mind, we designed into all airMAX AC Gen2 hardware (more on this later) an additional dedicated U-Mobile management radio allowing for a direct connection to any smartphone or tablet via wifi, greatly improving the deployment experience.
Network Analysis — The New AirLink
AirLink has been successful as a simple point-to-point performance estimator, but in order to make it a powerful tool WISP’s can use to better understand their existing networks, we are introducing terrain signal mapping for multipoint coverage analysis to greatly reduce the need for in person site surveys. AirLink will also have an increasingly important role in our future WISP technology roadmap.
Network Monitoring and Management — The New AirControl
Yes, AirControl has a history of stumbling. The original AirControl had a well-received WebUI, but was quite sluggish. The upgraded AirControl2 was very fast and scalable, but required Java to the dismay of a large part of the market. The “cloud” aspirations of AirCRM was poorly architected. We finally took a step back, came up with a final plan and have executed very well in the past couple years.
The new AirControl (v2.x to be exact) has taken the best pieces of the past experiences. It has a powerful and scalable back-end (like AirControl2), an elegant WebUI foundation with rapid feature development (much like UniFi), and a native development path to optional cloud hosting.
Customer and Operations Management — The New UCRM
As with AirControl, UCRM development had some challenges, but in the past year, a new team has done a fantastic job in making it into an outstanding tool. UCRM offers tight levels of integration between Ubiquiti device management and customer management such as customer Internet access suspension for past due bills and advanced QoS controls. With nearly 1,000 WISP’s using UCRM in it’s early stages of development, we are hoping the solution helps to solve many of the challenges associated with scaling from a start-up WISP to a large service provider business.
New Generation 2 airMAX AC Hardware
Traditionally, we have designed airMAX technology for long-distance open-environment links. In this case, antenna gain is king as higher signal strength leads to higher modulation rates and increased capacity. This was the logic behind our hi-gain dishes, sectors, and integrated radio products. However, a strong focus of our “Generation 2” airMAX AC hardware development was urban deployments where signal containment (and noise isolation) were often more important than pure link gain.
The IsoStation is a die-cast metal enclosed airMAX AC Gen2 radio that comes standard with a 45deg 15dBi horn antenna that has phenomenal energy isolation performance and is designed specifically to perform well in the harshest of RF environments. The PrismStation takes the concept to the next level — bringing our Prism active RF-filtering (in addition to the extreme antenna energy isolation) performance to the radio. Either product can be used as either an AP or a Station and can be upgraded with a variety of aftermarket horn antennas with varying coverage patterns.
Urban vs. Rural
IsoStation & PrismStation
The PrismStation and IsoStation (along with all Gen2 hardware), contain an independent AirView spectral analysis radio as well as a dedicated wifi management radio allowing U-Mobile support from any tablet or smartphone. In addition, Gen2 hardware ESD protection has been significantly enhanced in attempt to eliminate ethernet field failures. Best of all, the receivers of Gen2 radios have been redesigned with hardware tricks we have learned allowing them to be further resilient to RF noise.
Finally, the LiteBeam and PowerBeam Gen2 hardware have significantly improved mechanical mounting.
Breakthrough airMAX Wireless Performance
Perhaps the most important focus of our recent airMAX development was improving the upgrade path of new networks from airMAX to airMAX AC. Although it has been a challenging past few years, we finally see the light at the end of the tunnel. Recent AirOS firmware advances (especially with V8.1), have shown terrific results in adding airMAX AC radios to existing airMAX networks. Combined with Generation 2 hardware improvements, airMAX AC is now consistently outperforming expectations.
Coming Up Next…
This deserves a kudo. Well done.
It was the time for you to concentrate on the WISP industry other than the UniFi and AmpliFi and their shiny little boxes.
Now waiting the LTU wire up …
Very nice - I've always believed that people can best learn about things starting with the history of what led up to them.
I really enjoyed reading your story. It is a great one that you are still writing. Ubiquiti products enable me to make better internet for many rural customers without any outside capital. I appreciate the products you have made because they have allowed me to create fantastic internet for myself, my neighbors, my community and surrounding communities. It has improved the quality of life of millions of people around the world. I am very excited to try out the new AirLink and the new Gen2 hardware that is on the way! I purchase, use and recommend Ubiquiti products because of their performance and their value. I am invested in the Ubiquiti company for the exact same reasons, but mostly because I believe in you, your vision, and your ability to execute on that vision using the model you have chosen. This is great work and thank you very much.
This packs a lot of content into one integrated story!
Congrats on your saga of perseverence. Dave
Robert great job, how about posting this on your blog and twitter am sure would be great for investors to see your vision
Wow. What a great post. I fondly remember the original PowerStation as well. :)
Great article and thanks for the humility expressed in this article. Sometimes, the best way that we can learn is through failure– all the while striving to be better. We all can empathize with the challenges and successes expressed here. Well done.
10 years ago Robert called and asked us to fly out to San Jose. He said simply, "I've got something you have to see." After seeing the new NanoStations, it became very clear that Ubiquiti would be like no other company we had worked with before.
Oh, and yes, a 5-unit master carton of PowerStation model PS2-18V weighed 24.1kg!
Our first NS2 is still in service, providing WiFi access to a friend's outbuilding.
I still have NS2's and Loco2's still in operation and use them as seed for new AP locations.
Does any one know if Ubiquiti is having any interests in production of TV White spaces devices?
Huge achievement and impressive history, Robert.
When I saw the very first "integrated" Nanostation in times we had to solder coaxial connectors on customers's roofs, I thought: Finally! Someone took the Canopy's dome type enclosure and made a complete cheap Wi-Fi system into it!
But, at that time, there was yet another unique feature to get inspired by Motorola's "timeless" WISP CPE design (beside the dome integration): the TDD time slot synchronization…
After more than 10 years, you are still missing it. Will you, please, put more effort on developing a functional airSync for recent airMAX devices? This, together with a reliable "carrier-grade" radio statistics per every CPE (which I believe were introduced in AC firmware), seem to me more important than any of the new features you are mentioning here.
I don't think you'll see this again. First, last I heard, UBNT never had trademark to "Airsync" (Proximetry did). So not sure if that name is still around in UBNT literature. Also, TDD time slot sync works best when there is no other interference or competitors around. So with every device having wifi these days, it will just cause more interference for the other devices and for itself because it can't co-ordinate with the non-UBNT devices. Also, to get really high throughput, you need to use all the available time, and not eat up space with time slots that need margin between time slots. Lastly, with standard features like Multi User that is able to actually make use of simultaneous transmissions, the reasons and ability to run TDD is diminished.
The problem with statistics are two fold.
Most people don't really understand them. They are different for different environments. Stats also don't always reflect the given outcome, hence why many products are very sparse on stats. Does a CPE retry because of interference at the AP or because the EVM on the CPE was poorly calibrated causing 10%+ retries even conducted (Yes, I've tested UBNT in lab before, so I know how quality is all over the place. Transmit power can vary by 13dBm, RSSI by 20dBm. Target transmit rates different at different modulations, etc).
Stats take up CPU and memory space. When the CPU and memory becomes multiple times faster than the maximum data it can push (so its not a bottleneck), then stats will get beefed up. Otherwise, you're wasting CPU and memory on things that are viewed mostly on demand or possibly never. It also requires adding overhead to your backhauls to get this reported information to an EMS. You also really only get the AP's stats, unless you want to add even more over the air overhead by getting CPE stats sent on a regular basis, causing more retries, etc.
p.s. Its really, really hard to make the hardware transmit and stop at the very specific slot times without crashing the driver or breaking the slot time (or adding more overhead margin). The hardware is not designed like WiMAX or LTE radios. You either have more data to fit in a slot and have trouble putting the brakes on (causing buffer issues) or you don't have enough data and you're wasting slot space. Again, this will get better as the hardware improves beyond the peak wireless speeds. Also, the AP needs to process a ton more data than the CPE's to handle the buffers and scheduling. There is a reason an LTE base station will be 100X+ the price of an LTE modem. They also have the spectrum all to theirselves, so they don't have interference and retries to worry about as wifi does in unlicensed band where you shouldn't be killing your neighbours wifi by violating backoffs and what not.
I believe stats are much better on recent Ubiquiti devices. We even stopped pinging each and every STA from NMS in 10 second intervals just to get at least basic measure of transmission quality. Most APs need manual channel change several times a month (because of competitor's interference), but it is much easier to detect the need now. I even believe Ubiquity will automate this "manual" work in near future (Airmagic + action rules).
Concerning airSync. Will Ubiquiti officially abandon it for devices based on Wi-Fi chips? Wasn't it the claimed hardware difference of M (AirMAX) series (as compared to "bare" Atheros radios in … Mikrotik) what should have allowed for timeslot synchronization? I believe it is hard if not impossible to do this in firmware only, although Wi-Fi is much more aware of precise timing than most people imagine (prepared for PCF etc.) but I also trusted the information from Ubiquiti their hardware is ready for airSync... Should the airSync have been available (affordable) few years ago, we wouldn't have to fight even competitor's interference now: there are only few (3-4) dominant WISPs in our area - using "global" airSync, we could have settled a global coordination, at least at some channels.
Here in the Czech Republic, I have seen a TV tower leased by many ISPs with several tens (50?) of P2P dishes in total, all sharing a specific (Czech) band with 4-5 non-overlapping FDD channels. With TDD airSync, in theory, we could have achieved the same in the 5 GHz band (well, a bad example as airSync is more designated to P2MP networks).
You must be new to being a WISP. The more interference you have, the more you need sync. I have over a dozen WISP competitors, in addition to residential, private commercial and government users.
He who collects all of the data can make the best decisions about his network (or the automated software that feeds off of it).
How does one sync an AP to a competitor's AP, which doesn't have TDD? How does that work when listen-before-talk is disabled to force transmit whether a neighbour is transmitting or not? You need to have the RF parameters in sync, or else they talk over each other…
You can only control sync with your own devices. All TDD sync does is cut down on retransmissions due to collisions from multiple clients (hidden node) and make it fair, so that closer clients don't win over further clients, hogging bandwidth. The higher the client count, the higher overhead and decrease in efficiency, which is what TDD improves upon. There are tradeoffs with TDD, which makes it unsuitable for high bandwidth-low device count networks (ie, backhauls/ptp's). TDD is intended for increasing cell counts. In order to sync with an outside device, you need accurate GPS timing and an agreed upon transmission. GPS sync can be problematic if there is GPS interference or obstructions preventing a proper fix. Some other solutions use wired sync, which limits the co-location sync to one tower, which is often good enough if the next tower is km's away.
Anyways, unless TDD became an industry standard, different MFG's can't play nice with others, but there is limited spectrum so there is no reason to implement things that make your device transmit less often. It becomes a competitive advantage when your device transmits no matter what (which is not legal in all regions).
As always, highly directional antennas and high rejection receiver filtering is the key to a well operating system.
I don't know what plans Ubiquiti has for their TDD stuff. All I know is another company had that name trademarked and Ubiquiti was supposed to stop using the name. I think they just call that airMax now, no?
I also know of the challenges for Wifi spectrum and chipsets to do this stuff, and its a PITA. Many operators do not understand the preplanning and interruptions involved with running a TDD operation, whereas most small guys just want to be able to put up any AP in a day and just have it work. I don't see wifi TDD going past 11n, due to all the improvements in 11ac wave 2 and 3.
You share your sync settings with your competitor. It happens all the time. AFAIK, all of the Cnaopy operators around here are on the same settings. I have the same settings on my ePMP gear as the other ePMP operator in the area. It's a mutual benefit\mutual destruction scenario.
I don't think you know what GPS sync does. All APs on all towers talk at the same time. All APs on all towers listen at the same time. Obviously the CPEs are the reverse of this. When this happens, an AP isn't talking when it's neighboring AP is listening. The only sync that should be under consideration is GPS sync. Mobile wireless uses GPS sync. Canopy uses GPS sync. ePMP uses GPS sync. AFx uses GPS sync. LTU will use GPS sync. Mimosa uses GPS sync. Radwin uses GPS sync. There is more mobile wireless than fixed wireless and there is likely more GPS-capable fixed wireless than there is not-GPS-capable fixed wireless.
The problems with GPS and sync in general just aren't practical concerns. It's like shopping for a sports car and then complaining about the lack of towing capacity. It's true, but irrelevant. If they were valid concerns, the mobile wireless industry woud be in shambles.
You sound like a college graduate with no experience. He knows what the book tells him and he knows that theory very well, but he hasn't spent a day in real life actually doing this yet.