Tales from the Towers, Chapter 41: If I Had a Hammer

In this article, Rory Conaway writes about how wireless equipment manufacturers should be pushing the boundaries of technology, not just giving WISPs what WISPs think they need today. Doing so would lower dramatically the cost of deployment and allows WISPs to compete with wireline broadband providers.

Every time a new product comes out, there is a lot of discussion about pros and cons, and how it works within existing deployments. The primary manufacturers have been following the same formulas for years. Cambium Networks (a spin-off from Motorola Canopy) targets high-density, long-range tower deployments with its tower-centric hardware product line. Ubiquiti’s product line is pretty much designed for everything else that is deployed by wireless ISPs (WISPs). This long-range tower-centric model is the only one that makes sense from a financial position for WISPs with Cambium equipment. When an access point (AP) costs $3000-$4000 or when it costs $10,000-$50,000 to outfit a tower with APs, routers, backhaul radios, and battery backups, it’s imperative to maximize the efficiency of that tower to increase profitability.

Although Ubiquiti can do tower-centric designs, long-range works best in areas where interference is at a minimum and line-of-sight (LOS) deployments are feasible. It’s a challenge to make the 802.11n WiFi based protocol work in non-line of sight (NLOS) environments and the lack of GPS synchronization limits high-density AP type deployments. That means towers aren’t the best in cities and suburbs as I’ve covered in many Tales of the Towers articles. However, it’s more flexible and cheaper with the AP/CPE option built into a wide variety of products, some of which start at less than $50 (let me introduce you to my little friend, the Ubiquiti NS2M Loco).  In high-density areas though, I’m moving everything to 5GHz which means the NS5M Loco for $20 more.

What is more interesting to me is that all of the Motorola Canopy CPE product lines look the same with an integrated radio and relatively low-gain antenna. Even Ubiquiti borrowed that same design, albeit with electronically switchable dual-polarity in their Nanostation product line. Move forward to Cambium’s 450 series, and they are sticking with the same formula. Put up the AP, if you need more range, add a reflector. Why mess with success, right? Let’s just say that my enthusiasm for that design today falls somewhere between a teeth cleaning and a root canal. You need it for Cambium because that’s the only real option they give you, unless you want to connectorize everything. But what if we start over and really look at what works best for installers and my Diet Coke fund (I keep trying but I’m not getting product placement endorsements yet)?

Ubiquiti opened up the drawing board a little further. The Bullets were the first radio that broke the mold and went right to solving a problem. The entire AP, screwing right into the back of the antenna, was innovative and clever. The NanoBridge product line was next, unique in both cost and performance. You can’t beat 25dBi of antenna gain and a lower profile than a reflector for less than $100. What a deal! That means J-mounts without reinforcement (although finding a 30-inch J-mount that isn’t back-ordered for 2 months has been a challenge),  lower wind profile than a reflector, and it’s a cheaper install even if it didn’t cost less. MikroTik does the same thing but the gain of their antenna barely trips my gain meter. The MikroTik unit is cute though, kind of like a little porcelain knick-knack that my grandma puts on her fireplace and doesn’t really do much.

But time marches on and so does DOCSIS 3.0 and VDSL. I suspect VDSL is probably the end of the line for twisted pair so that just leaves cable and fiber long term. Intel is pushing their new 1Gbps DOCSIS specification, but I have to imagine that means RG-59 is going to become the elephant in the room, so to speak, that is going to all have to be replaced and the cable companies aren’t going to pay for that. The problem with all of us is the assumption that our way is the right way forever. If you are profitable and still growing without sacrificing the quality of service with your existing clients, then it’s the right way now for you. However, as WISPs start to overlap each other and run up against Wireline providers, the challenge to continue following the same old ways when you are totally aware of limitations becomes pointless. And if cows could fly, I’d never have known the joys of chocolate milk.

Manufacturers have unknowingly artificially limited us in these areas because they are building what they know. Ubiquiti’s dual-function AP/CPE M series product line (yes, I know they weren’t the first but they definitely took it to the market successfully) has let me do things I could never do with Canopy. Guerilla Wireless and S.P.I.R.I.T are two of the models I’m deploying that would never have been feasible with a rigid tower-centric product line.  AirFiber is another product that will help me enhance these models.  As each new tool comes out, more possibilities occur. However, these incremental changes aren’t sufficient to really take on the cable providers yet and maximize profitability. At the rate the wireless industry is moving in this area, we will be competing against DOCSIS 8.0 by the time we get there.

A lot of the manufacturers’ decisions on what to build are driven by us. They build what we tell them, what we know and consider safe and comfortable, sometimes to our own detriment. Check out the flying wing concept plane Boeing wanted to build. It would have revolutionized the cost model for the airline industry years ago. However, researchers found that passengers wouldn’t fly on them without windows (boy, I wish I were there to smack the people who insisted on this the focus groups, dooming us to 100-year old designs), thus airlines weren’t going to buy them, and hence it died. Instead we get the AirBus and the Boeing 787 Dreamliner.  Yes, they are better than current models but nowhere near as good as we could possibly get.  We get evolutionary, not revolutionary (okay, except in the case of AirFiber and innovation from SAF and Trango to compete). Fortunately for us, the customers could give a flying donut as to what our APs look like as long as their house doesn’t resemble Mission Control. The exception to that rule was my Grandparents who let me put up a 40-foot telescoping mast with a 22-foot Avanti Sigma 5/8 antenna on their roof with guy wires every 10 feet.  Seventy feet from ground level meant that as CB base stations went, it was very, very cool.  Who knew that nylon wasn’t a good guy wire?

If the wireless industry wants to really compete with the next frontier within the limitations that the government has put on us, we need to start over. We need to put on our big boy pants and look at the WISP industry holistically, not specifically.  We have radios, we have network components, then we have accessories, and finally we have existing deployment models. Now that the end user wants NetFlix and DOCSIS 3.0/VDSL are happy to deliver it, we need to fundamentally start over and design for that and what’s beyond that. For Wireline, it’s fiber. Based on what I’m seeing with our economy and our wonderful share-the-wealth president trying to suck every penny he can out of the free-market in taxes, (yea, that makes sense), there will be little funding available for fiber in the near future.  Wireless, with the right deployment model can fill that role, especially with what we have coming next.

Ahh, a pipe dream you say.  Nay, Nay I respond.  Remote WISP deployments aside, think about how you would deploy in every neighborhood in the country if you hadn’t ever seen a Canopy or Nanostation product line. Shake the Etch-a-Sketch and start over with your magical 3-D Printer which could create whatever you could dream.

The next component in this process is a manufacturer that is willing to start over. Of course what equipment we have out there is great or none of us would be here.  However, we are still hiding in the woods and behind the hills. Very few of us are directly competing with Wireline from day one. If you have 5000 remote clients paying $50 or more per month, you can subsidize growth into an area that might have a lower starting revenue point. However, if you are starting from scratch tomorrow, even if you are sitting with $1M in the bank, how do you go right into a Wireline area, deliver the quality of service that Wireline providers provide along with their speeds and expectations and compete with $20 per month? I’m doing that now by designing around what I have available instead of defining the standard of what I should be doing. There is lot of opportunity that we are still leaving on the table.

For that, we are going to need some manufacturing help on several levels. The problem is that building the type of system we need to compete with Wireline, will require a new paradigm and partnerships.  We don’t need any new technologies to build these type of products (although MU-MIMO will really help), we just need a different starting point. However, it will either take a significant investment or a lot of different partners coordinated in building this product line.  Third party products have been filling in many holes that manufacturers have left open for WISPs. Beehive created reflectors, surge protectors, and many other products. RF Armor built shields. MikroTik built routers. Tycon built PoE switches. What I’m seeing now is that manufacturers are starting to overlap and eat into each other’s market share as they expand their product lines.

So what type of product line is necessary to go from the 5% of the country that we fit into and get into the other 95%?  Mesh obviously isn’t the answer although it was a good first effort. The original products were too expensive to deploy and maintain, and were unable to scale. Mesh itself wasn’t the problem, the deployment model of mesh also held it back along with the routing protocol.  Then add in technology obsolescence which meant upgrading was going to require the complete replacement of every single piece of equipment. When you have spent at least $100,000 per square mile to deploy a full coverage system, that’s not even realistically financially possible. That and they ended up being far slower than their marketing material implied (yes, we all use UDP for all traffic so let’s put that in the headlines instead of TCP/IP at half the speed).

PTMP tower based systems clearly aren’t moving into New York City and with the limitations placed on WISPs for free frequency bands, they aren’t going through buildings and trees. Throw in the interference levels now being created in all the bands from WISPs, personal use of wireless equipment, and SCADA use in the 900MHz. White Space hasn’t developed where it needs to, but it will be pretty much useless in or near most major cities.

That’s the dire reality.  However, engineers are a very resourceful bunch. Ubiquiti’s AirFiber as a (point-to-multipoint) PTMP product is really cool not just what it can do today but what it can possibly do down the road.  I’m not talking about simply increasing the throughput with a higher QAM rate, but using it as the foundation for a whole new product line that isn’t based on 802.11. No, I don’t have any information that hasn’t been all over the forums or casually discussed at WISPA, but it is a whole new chipset from scratch. It’s also one that split the transmit and receive into 2 different antennas. It’s not hard to extrapolate what that gives Ubiquiti in the toolbox down the road, including a proprietary RF technology.

Although little MikroTik routers based on Atheros and Atom processors are cool, Xirrus radios using Cavium processors (meaning really fast and cheap) brought a whole bunch of other possibilities rushing to the front of my brain. Then Ubiquiti announces stacks of routers that are either based on those or something very similar at prices almost in MikroTik’s sweet spot but with more power than a steroid laced Barry Bonds. However, Mikrotik  isn’t going quietly and has introduced its own router line that also looks like it’s based on the same router core at the upper end. The gloves are off and I love a good free-market street fight. However, I go back to Xirrus again and really like the use of the Cavium processor simply in the AP. Clearly it’s being underutilized in their product since I still have to add more layer 3 processing in a large deployment. Also the APs aren’t exactly sold at K-Mart Blue Light special pricing but it’s a start.

That barely scratches the surface of what WISPS need to get into the alley ways and kick some butt. I have my own ideas about what’s needed and they are evolving almost daily. I threw the down the gauntlet a few months ago with S.P.I.R.I.T.  It was designed within the limitation of existing technologies at the time. However, many things have changed since then. One proposal we submitted with Cielo Systems a year ago was based on S.P.I.R.I.T. and cost about $10M to cover about 40 square miles.  As I see new products come out, I’ve been updating the spreadsheet to reflect that. If I were doing the same thing today, the cost would be down to about $6M with a massive security and mobile infrastructure geared around a municipal system. Take that out and make it a WISP model only and the cost drops to about $4M, or $3M plus residential truck rolls depending on where you put the Capex. I suspect that with some additional work, I could get it down to about $2.5M now. The difference is that the available capacity is now about 200Mbps per AP location versus our original 100Mbps. However, if this system got to redesign all the manufacturing equipment from scratch, integrated known and future technologies, including handling interference issues, the cost would drop to a deployed cost less than $50K per square mile and would have fiber capacities per AP in terms of speed.

I’m not saying that $50K per square mile is still financially feasible as a pure profit play (as an average cost), but that means some areas could be a lot less, others more. I can see getting the cost down to less than $2500 per square mile in many situations (Arizona is a good example). Just imagine if you can deliver 50Mbps per house consistently at Ubiquiti prices and how that affects your ROI and ability to compete. I’d like to see that someday but at the pace that the industry is currently getting there, I might be too old to climb a ladder when it does. It’s time to start over and grab the Legos, we have some building to do. Even worse, with Hostess filing to go out of business, we need to do it before my supply of Twinkies and HoHos runs out.

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.

Speak Your Mind