Before you start reading this article, let me warn you first that it is longest one I’ve written and the only one in which the manufacturers I talk about see the text before publication. I have my opinions and perceptions as many of you do. However, I also wanted this article to be factually accurate and gave both manufacturers a couple of days to respond to any inaccuracies. Designing a WISP with a tower-centric model is pretty straight forward. Many of you have done that successfully for several years. In the beginning of our industry, that was pretty much the model that stands proven and true today.
However, today attempts to improve upon the model range from a friend who wants to cover his neighborhood to companies that are looking at going public down the road. As Open-Range, Key-On, and several other recent failures have demonstrated, that type of model is more difficult to design and deploy successfully (read profitably) in today’s competitive environment in the United States. As the United States is only 20% of the worldwide wireless ISP (WISP) market, I’m not going to presume that my analysis applies everywhere. In two other countries I’m dealing with, both models are valid and I’m looking at hybrid models in each of them.
As recently as 3 years ago, I saw areas with hundreds or thousands of users that still couldn’t get any high-speed wired or cellular-based broadband. I’m not counting the first generation satellite systems since they could get everywhere, but were very limited in capacity. The typical areas that still don’t have coverage are not the small pockets of dense housing areas but simply areas that are remote, where houses are few and far between.
However, throw in politicians who still haven’t figured out that stealing money from taxpayers and giving it to their big donors is reprehensible. Even worse, they couch it in terms of “helping the poor. “ If you are going to use taxpayer money, $300,000,000 of it for example and give it to the 5 biggest cellular/ISPs in the country to help them wipe out small, innovative competitors in the free market system, at least be honest with the public. I’d have more respect for a politician who simply says that he has to repay his biggest contributors who lined his pockets during election time than to use the poor as an excuse. With our brilliant politicians and bureaucrats, what’s remote today might become a subsidized competitive environment tomorrow. For example, we started launching in one area on the exact same day that Comcast and Qwest publicly announced they were coming into the area which has made for a great test environment. There are still many areas that have limited coverage but they are getting harder to discover. We need to find ways to be profitable in small areas and suburban America to be successful in the future.
The problem with these smaller areas is that in a tower-based model, it is financially more difficult to recoup a high capital expenditure (capex) because of the low density of the population. To become profitable, the old model of high-capex tower deployments with long-term clients has to be reassessed. Since I’ve never been one to be politically correct, I’m going to compare the models of the two biggest vendors, Cambium and Ubiquiti to discuss some of my design criteria in this area. You be the judge as to which one works better for you as a WISP. For full disclosure, I don’t operate a Canopy-Based WISP nor do I have stock in Ubiquiti. I’m a Cambium, Motorola, and Ubiquiti value-added reseller (VAR) and have deployed equipment from all these vendors. I’ve also engineered Cambium/Motorola deployments in the Public Safety market. I deploy what makes sense from a financial/quality/performance standpoint and wouldn’t hesitate to deploy equipment from a different manufacturer if it the numbers work. Unfortunately, few investors today are interested in a 2-year profit point and 42+ month ROI. The pressure is very high to turn things around quickly when you are playing with other people’s money (OPM).
Motorola’s Canopy/Cambium: Solid engineering, difficult and expensive to deploy but suitable for certain conditions
While Motorola was selling off its cellular division with all its patents to Google, it also sold off the point-to-point (PTP) and point-to-multipoint (PTMP) products to an investment group, Vector Capital. This line of business (PTP and PTMP products) became a separate company called Cambium. The fact that Motorola diversified their company to this extent is more of a demonstration of tired, old engineering management at Motorola that lost its innovative spark and foresight starting 20 years ago than any great business move. Any innovation within the companies seemed to be squashed from management who lost sight of the customers and markets. They were too busy congratulating themselves and thinking nobody could compete with them. It eventually became about the quarterly numbers and bonus checks, not about leading the industry. New technologies had to be brought in from outside and then they typically died with little further development or innovation. Ubiquiti provided a very rude wake-up call in that area.
Although Orthogon and Mesh Networks are good examples of companies acquired by Motorola to complete a product line, I think the best example of lack of development and foresight is Canopy. A brilliant product and product line designed completely in-house by engineers with a passion. It was the best of Motorola engineering when that mattered 11 years ago. The problem was that it was never done with the company’s blessing and had to be driven from engineers on their own time. The Canopy product line became the most reliable and solid product on the planet for WISPs. Then it languished for development money for the next several years while committees, accountants, and management ran it into the ground. Instead of advancing the product line, they created newer, incompatible products. Canopy is still the most reliable with “mean time between failures” (MTBF) ratings measured in decades, not years. Unfortunately, you can’t design a product 10 years ago in a high-tech society and expect that product to stay at the top. Although solid beats flash-in-the-pan, over the long term, NetFlix, YouTube, VDSL, and DOCSIS, and FTTH meant the Canopy 100 series is going to be heading into retirement in many countries.
Keep in mind that end-user bandwidth is not as important in other areas of the world. For example, in countries such as Afghanistan and Nigeria, bandwidth could cost $400-$1000 for a 1Mbps circuit. In those markets, and especially the poorer areas, 256Kbps is a pretty good rate and a Canopy 100 can still be successful if the equipment costs can be kept down. However, capex is still a problem since the companies attempting to launch systems can’t charge $10 per month for service and spend $15K-$60K in equipment on a tower and $300+ for an installed CPE. This works if the government is funding the infrastructure but it makes life difficult for entrepreneurs to get started. However, if you are the government and you want a rock-solid tower-centric system with a company that can hold your hand from start to finish, you should consider Canopy. The equipment covers every band from 900MHz to 5.8GHz and will last longer than a Twinkie (a Hostess snack designed to be edible for at least 50 years in my opinion and second only to the Hostess HoHo in awesomeness).
Of course Cambium has several other Canopy PTMP products such as the 320, the 430, and the delayed 450 product line. However, except for the 450 which is being rumored to cover other bands other than 5.8GHz in the future, they are all focused to a specific band. They don’t cover all the unlicensed bands like the Canopy 100 product line. Although the 450 may bring back that diversity someday, it’s still months away for the first band and many months past that for the rest. As each of the new product lines were built, they had one thing in common: they were designed by engineers for engineers and did not benefit from new user human engineering and friendliness. Even worse, there was no common design between them, no compatibility, no upgrade options, and training started from ground zero. That meant that the cost of getting into the market had to include several days of training or some really good experienced friends or consultants. Nobody buys a Canopy 320 to play with before they decide to get into the industry.
Ubiquiti: off-the-shelf wireless network to go, the WISP’s best friend
On the other side of the equation is Ubiquiti. Take off-the-shelf chipsets, mix with an open-source operating system, and customize it, package it like a Ferrari, price it like a Yugo, upgrade it every couple of years with new hardware but keep it compatible and with similar firmware GUIs and settings, and make it user friendly enough so that every Windows user on the planet can set up a WISP. MTBF isn’t anywhere near as important if the client radio is a 3-5 year throwaway. Where Cambium supplied factory support trained technicians, Ubiquiti supplied a user bulletin board. Where Cambium has channel and dealer representatives, licensing contracts, and distributor incentive programs, Ubiquiti had Ben Moore and an inventory search option on their website. Instead of R&D teams, market analysis reports, manufacturing costing, middle management reports, quality control processes, Ubiquiti replaced them with Robert Pera doing all the hiring, and the middle and top level management decisions. They also used standard chipsets, outsourced engineering, customer technical support, product licensing tracking, overseas software development, and pretty much the user base as your quality control division, products are simply cheaper to make. Ubiquiti also doesn’t have business development managers, human resource personnel, and many other layers of management like Motorola did. Since the IPO, Ubiquiti has added several Level 1 support staff and channel sales personnel but product availability still seems to be an issue. Simultaneously, Cambium seems to be putting the message out that they no longer have those constraints and that we should expect to see a faster, more customer responsive company.
Of course, you can announce that you have achieved Faster-Than-Light Speed by rubbing magic gel all over your body but unless you can produce a product and actually deliver it when your customers need it, it’s irrelevant. GPS, AirBeam, and cable that doesn’t fall apart in the sun after 3 months means that rushing to market and bypassing some engineering Quality Control process is not always a good thing. Many WISPs who deployed early equipment found out that saving a few bucks up front gets a whole lot more expensive when they had to go back and replace hundreds of cable runs at residential houses, rent $1200 per day lifts to replace radios that lasted a whopping 60 days on a tower, or had to drive out hundreds of miles to apologize and explain to customers who woke up off-line because firmware version 5.x.x wasn’t tested for all the bugs before 16 beta releases. To add insult to injury, many WISPs committed to deployments and a schedule when all of a sudden, products were impossible to buy through the channels and installations got delayed. When you are the underdog, you can get away with that but when you become a market leader, continuous product failures and vaporware promises put you in the same position in which previous market leaders were and make you highly vulnerable. The best specifications are also useless if the product requires a lot of expensive handholding to keep operations running, and if it can’t scale into large area deployments, and can’t be sold because product simply isn’t available.
So with all these issues, why has Ubiquiti become so successful so quickly? Simply put, Ubiquiti made it financially possible for thousands of new WISP operators worldwide to come into the industry. I’ve seen WISPs now start with less than $10K. Ubiquiti makes it easy to set up small area WISPs cheaply and quickly. The broad spectrum of product line now matches the Canopy 100 and the user friendly web interfaces cut the learning curve way down. New users can buy a couple of Nanostations for $100, teach themselves how to make them work in minutes, and set up a 20 user WISP in the blink of an eye. The recent growth spurt of WISPA is directly related to Ubiquiti and is helping the industry gain more lobbying clout. If you run into a few bumps along the way due to product quality control, compared to not being able to get into the industry at all, it’s the gamble you take. I’ve seen significant improvements in many areas but there is still a long way to go in many areas.
I also don’t’ see any other good options at that price point (boy, am I going to get hammered on this one and yes I’m aware of Arc Wireless, MikroTik, and a bunch of other vendors) that directly compete with Ubiquiti at the low end. None of them has the breadth of product line at this point so if you want one throat to choke, then you understand. When switches and routers start shipping, the throat is going to get a lot bigger. Along with some really clever mechanical and antenna engineering, Ubiquiti firmware allows every radio to either be an AP, CPE, or both. For $300, you have a 100Mbps+ (aggregate) dual-polarity 360 degree access point that can easily cover 4-16 square miles. That’s simply a no-brainer for low-density areas. In addition, the future promise of functional GPS and current semi-functional polling still makes it better than standard 802.11 devices with neither option on the drawing board. The management tools are also extremely functional and free in case that got missed.
Differences between Canopy and Ubiquiti
However, everything is a tradeoff and the decision to deploy either product really needs to be thought through very carefully by WISPs. At present, I see 2 distinct differences in the product lines of the manufacturers. All of the Canopy products, the 100, 320, 430, and 450 series are completely designed as tower-based systems. With a single exception in the 100 series, these are straight-up Access Point, CPE products. They are engineered to operate at very long ranges consistently and dependently. Since towers are built for large coverage areas, features such as Polling, GPS timing, and Frequency Reuse maximize efficiency within these product lines. GPS synchronization also minimizes interference between towers and with cooperation, even between competitors. Polling means that as the APs handle more clients evenly, CPEs are still given each time access and don’t lose efficiency to the hidden node interference problem that plagues 802.11 a/b/g/n systems without it. If you have problems with your setup, Cambium is also there with the technical resources to make that work. There is something to be said for solid engineering and the resulting lifetime performance.
Ubiquiti had the advantage of industry wide standardized communication protocols behind their products. That means state-of-the-art 802.11N 2×2 MIMO protocols with real-world throughput capability of up to 300Mbps. However, the protocol was never designed to be installed on a tower with hundreds or thousands of simultaneous users and at long ranges. It was designed to connect your laptop to AP in the next room. GPS synchronization, polling, automotive power adjustment, and many other outdoor based ideas were never considered when committee after committee ratified the specification. Ubiquiti took the specification and extended it into a new region, the outdoor arena, first with 802.11 a/b/g radios and then into 802.11N. However, when 802.11N 2×2 MIMO came out, the game changed. Where a Canopy 100 AP can handle up to 14Mbps aggregate, a Ubiquiti M series radio can handle up to 80 Mbps. Keep in mind I’m comparing product lines that cover 900MHz to 5.8GHz, not individual products like the 320 or 430. The Canopy 100 series is a lot closer in price point. People were drooling to look at this technology as cracking into competition with wireline services.
Some of the perceived problems with Ubiquiti came about simply because the product line didn’t fit into the model of the tower-centric model that everyone was used to, at least early on. Early firmware attempts at polling with limited processor speeds, even in the M series, meant that the 80Mbps AP is going to be limited to about 40Mbps with significantly higher latency as the number of users scale up. GPS features have also turned out to be a big bust with limited success in 5GHz and no success in 900MHz or 2.4GHz. Without these features functioning, frequency re-use is not even an option on most towers. Some third-party components that help solve the problem of poor-shielding in the Rockets and antenna leakage such as RF ARMOR makes it possible in the 5GHz band. These products do help with noise in the antennas anyway, but it’s no substitute for these features in the base product. Ubiquiti’s response to this is the next generation Titanium product at 2.5 times the price which like the 450, hasn’t shipped yet as of the date of this article.
What Ubiquiti is good at is a low-price product that is the Lego of wireless. Models can be modified to any environment. Guerilla Wireless and the Secured Public Integrated Radio Infrastructure Technology (S.P.I.R.I.T.) models both demonstrate different ways to make the sum of the parts perform far greater than the whole. Ubiquiti can go from Backhaul to PTP to Pseudo Mesh depending on how you put the parts together. In every case, the cost/performance ratio is 3-10 times better than industry standards.
Even with all these issues, Ubiquiti has been a monumental success that has changed the WISP industry. In a few years, they have overtaken Canopy sales and have become the darling of almost all the new WISP operators and many of the more established operations. All doubts about this were erased when at the WISPA convention in Las Vegas last October 2011, the Cambium breakout rooms had tens of operators and the Ubiquiti breakout rooms had several hundreds more.
The reality though is that Ubiquiti isn’t a plug and play replacement for Cambium. You can’t take down a Canopy 100, 320, or 430 and throw up a Rocket with a sector antenna in 900MHz or even 2.4GHz on a tower with 600 users. Ubiquiti wasn’t designed for that yet. Without GPS, frequency reuse is almost impossible except in certain circumstances. You also can’t play in the same box as your competitors if they don’t share shovel and the bucket willingly at pre-determined times. When you can use a water tank or a building as a shield, then you have more options. In 5.8GHz, the RF Armor shields works pretty well but because you can do it doesn’t mean it’s the best way in all situations. At lower frequencies, it’s not as good but it also adds wind load, a killer on towers. A lack of a true outdoor type of RF protocol also means interference is more bothersome and it takes more time to manually adjust radios with power settings to keep things stable.
At this point, I design with the idea that Cambium still owns the towers for high-density, long range deployments and Ubiquiti is better served for deployment in smaller area coverage zones with fewer users. Instead of one tower 5 miles away, use five buildings that are 1 mile away (yes, that doesn’t scale exactly but you get the idea). Since Nanobridges can easily go 2 miles and still deliver 100Mbps or more aggregate for less than $200 for backhaul, this is a whole lot cheaper. It’s slightly more work but it’s very inexpensive to trade roof rights for internet bandwidth. Even if you need a telephone pole and electrical, it’s still cheaper to put one up and pay a monthly electrical bill than lease tower space. What’s good for the industry right now is that we have never had so many good options and the entrepreneurs are jumping into the business at the fastest rate ever because of it. In part 2, I’ll cover what the manufacturers are currently up to and how it affects the future.
S.P.I.R.I.T. Follow-up – Since the original design in Evansville, Indiana, S.P.I.R.I.T. has been undergoing a design change to incorporate new technologies. With these changes, the design can now easily support fibers speed capacities down the street. Simultaneously, the cost has dropped probably 60% from our original estimates. These changes will be incorporated into our online configuration tool when it’s completed.
* * * * * * *
About the author
Rory Conaway is president and CEO of Triad Wireless, an engineering and design firm in Phoenix, Arizona. Triad Wireless specializes in unique RF data and network designs for municipalities, public safety and educational campuses. E-mail comments to rconaway at triadwireless.net. Rory writes regularly for MuniWireless.com.