Chapter 8 – Half of Something is Better than Nothing, but Why Settle?

Our network is pretty inexpensive but suffers from first generation bandwidth problems of losing ½ the bandwidth capacity every hop. Patton wouldn’t settle for a partial victory and neither will we since we can do it for very little additional cost. It’s possible to virtually eliminate bandwidth loss per AP for up to $400 for the Gateway points ($100 for every direction it has to backhaul) $200 for midpoints and $100 for the end point. Keep in mind you don’t need to do this at every AP, just the links that have a higher load or to extend the network further. This is also something that can easily be added later as capacity increases.
We currently have the AP cost at about $200 and each AP can handle 30-50Mbps of traffic. However, 2 hops down from the gateway point or backhaul point, we are now down to about ½ of that. Go another hop and its ¼ of that using single radio WDS functionality. WDS has 3 other problems that we can’t get around with Ubiquiti’s Bullet 2M HP’s. The first is that we lose the client isolation function. The second is we lose a proprietary function that we will later cover called AirMax. The third is that we lose advance WPA encryption methods and we are limited to the evermore useless WEP. It’s easy to get around this though. Keep in mind the motto, “it’s not a problem if it can be fixed with money”.
Ubiquiti sells a little radio called a Nanostation Loco M5. It’s available in both 2.4GHz and 5.8GHz versions. We will focus on the 5.8GHz version since that limits the interference for our backhaul and as we discuss later, opens up new options for our network for PTMP. This little fellow only costs $70 and provides a 2x2 MIMO signal that can support up a total of around 100Mbps of total throughput with a 20MHz channel. I’m going a little conservative on throughput numbers since different types of IP traffic can test between 60-150Mbps. Put 2 of them on the pole with the Bullet M2 and your backhaul can now support all that bandwidth across several hops with a loss of about 1Mbps and about 2-3ms of latency added on each hop. However, you have just added another $200 per AP to do this with a switch. If you have two 5.8GHz radios on the pole along with the Bullet, you will need a switch.
The most cost effective switch I have found that is designed to handle reasonable temperatures is the Linksys SD205 unit. Although they are rated for 122 degrees, I have never seen one fail and I have had them in temperatures far in excess of 140 degrees for years. For around $25, they are the best little units for sealed outdoor installations and there is an SD-208 version if you need more ports. Compared to an industrial switch which starts at $250 (cheapest ones I have seen), the Linksys units are a steal. If you need a managed switch, then you are looking at industrial switches and the cost is going to start around $450.
The other good part of adding the Nanostation Loco M5 to our system is now if we have to add more AP’s in the middle of our network, we don’t have worry about the 1/n issue. We can also use WPA or AES security on the 2.4GHz network and 5.8GHz backhaul hops. This solves the wireless security issue. If we have to go around corners or Y off an AP, we can add up to 4 Nanostation Loco M5’s per pole if we use a 20MHz wide channel. Yes, I know that theoretically you can do 5 channels but that leaves no buffer between channels. With 4 channels, you get a 5MHz buffer which although not ideal, it’s not bad.
The drawback to the Nanostation Loco M5 is that to get maximum throughput or MCS (15) rates, you will be limited to 17dBm output with a 13dBm antenna or a total of 30dBm EIRP. If there are any tree obstructions or you need more power, the big brother, the Nanostation M5 has a 16dbi antenna and 21dBm of output for an extra $20. Keep in mind that these radios will go to a much higher power level if necessary but it comes at a reduction in throughput which we discussed in article 2. It does however, leave some room for increasing power if interference or obstructions start to cause issues later.
The Nanostation Loco M5 radios can also do one other thing. If we get back to the idea of a PTMP hybrid system, that means that users that are within the beam pattern of the Nanostation Loco M5’s can actually connect directly to the 5.8GHz backhaul. I suggest this stay reserved for truck rolls and trained technicians. 5.8GHz is a lot more sensitive to LOS which makes it more difficult for clients to install themselves. We are going to extend this concept further later also.
If the goal is that clients are going to install their own CPE’s, either indoors or out, it’s best if clients use a 2.4GHz product, preferably a Nanostation 2M or the upcoming Nanostation Loco 2M for indoor use with a window/wall mount. There is also some new indoor equipment from Ubiquiti, the WiFiStation, with directional and omni-directional antennas that support 802.11N and only cost around $30. The range is obviously lower but it also lowers the cost for clients.
Although our network also doesn’t support 2x2 MIMO on 2.4GHz yet, it is now supporting it on 5.8GHz so things are moving along. We have a couple options coming upon the 2.4GHz 2x2 MIMO solution coming up. There is also a way to customize the firmware on these units so that the CPE can be fixed only on your network with the proper settings and would allow you access to manage settings on the client side to optimize their connection quality. Imagine the client connecting to the system and then you get to remotely upgrade firmware on every client device simultaneously, monitor their connection quality, and manage the connection all way to the computer. That is the advantage to having all the radios from a single vendor. Tech support calls would go way down and staff wouldn’t have to fudge their way through 100 different products. Taking this a step further, if all the devices were Ubiquiti M series radios, you could also use the AirMax feature with polling for the clients. Of course, you sort of kiss off the legacy devices, or do you?

Chapter 7 – One Size Does Not Fit All

Before we continue developing the system further, I think it’s a good idea to discuss all the various antenna designs that go into a municipal design process. I’ve described one type of design with omni-directional antennas, although I use many difference designs customized to the target client. It won’t meet all needs; no system will unless the budget is unlimited. It is being designed to be as flexible as possible, but there are specific technologies that may work better in some areas. Wireless hardware manufacturers have put forth various designs and optimized their hardware towards that goal. Budgets have sometimes dictated other designs such as “Tales from the Towers.” Due to new equipment that has been released, the numbers of potential designs that can be deployed have exploded.
Let’s start with the legacy systems that started the whole muni-wireless market. The most basic APs were single radios with a single, or dual, omnidirectional antenna design. These systems covered a fixed area and then simultaneously handled backhaul to the next radio. The first problem with this design was simply the fact bandwidth drops by ½ every hop. Although they also supported diversity for the antennas, that didn’t significantly increase the range. It did reduce fading since 802.11b wasn’t really created with a multipath environment in mind.
Eventually, manufacturers went to 802.11g and started adding 2-3 or more radios with secondary radios for backhaul. Using 5GHz frequencies to hand off the backhaul functions increased user throughput, and with dual 5GHz radios, the multi-hop bandwidth loss problem was effectively eliminated. The multi-radio design also spawned some extremely unique AP designs. Manufacturers started adding directional antennas, multiple frequency coverage, integrated sector antennas, beam-forming, and a few other ideas that don’t pass the smell test for actual performance advantages. However, the reason that there are so many systems out there is that there is an actual need for different features. I have used, and will continue to use, many of these products because of their uniqueness.
Jump forward to post-802.11N MIMO technologies, and the number of options from both technology and a budgetary position are mind-boggling. I typically go through no fewer than 5 designs and multiple products when trying to find the best design for a client. I still have questions on design ideas that I haven’t deployed yet that I’m testing. Since most APs today are multiple radios, the exception being the system we are designing for TriadLand, we are simply going to discuss the 2.4GHz side of the APs.
Because 802.11n is simply faster than b/g, we will stay focused there with the idea of backward compatibility. 1x1 MIMO, 2x2 MIMO, 2x3 MIMO, single-polarity, multi-polarity, and beam-forming are all being deployed. Which one is the best? Actually, most of them have some unique feature. It depends on the application. The right answer is the one that solves the problem within a specific budgetary or financial target. So does that mean there is a universal AP? The short answer is no. The really long answer, which I’m going to need 2 more pages to cover, is still no, but there are ways around most of the issues. Some of the answers, I really don’t know right now because I’m still testing some new ideas. Recent discussions and some new projects have gotten peaked my interest.
Let’s get back to our simplest AP design--our “Tales from the Towers” model. It’s a single omni-directional antenna on a 1x1 single stream 802.11 AP. If all clients are 802.11n compatible, then it can support 30-35 802.11b/g/n clients with a total throughput of about 30-50Mbps, TCP/IP. This assumes good LOS coverage, low-interference, and a low-reflectivity environment. We improved the range by using a very-high gain collinear antenna, which has a higher gain over most of the AP’s that use 6-9dBi antennas. It’s not perfect, but it’s cheap, and sometimes that’s all one needs.
802.11n has a distinct advantage over 802.11b/g, MIMO technology. To take true advantage of it, you need multiple either multiple antennas or multi-polarity dual-feed antennas. The question is, do you use multiple vertical antennas or multiple antennas in different polarities. Do you use 2x2 MIMO, 2x3 MIMO, or 3x3 MIMO? Which one works better with legacy 802.11b/g devices, and does it matter?
To answer that question, you first have to understand antenna polarity. The most commonly used polarities are vertical, horizontal, and circular. We used a vertical polarity omni-directional on our original design, which means vertically polarized in relation to the ground. We did it mainly for budget reasons, but how well it works depends on what the polarity of the client device is.
Let’s examine the typical laptop. Early WiFi-enabled laptops simply had small wires or circuit board antennas embedded on the WiFi card internally. Since the board/wire is typically laid flat in parallel to the table, the antenna would be considered a horizontal polarity antenna. So what happens when a vertical antenna connects to a client with a horizontal polarity antenna? The result is up to a 20dB loss of signal assuming both antennas on both sides are the exact same specification in alternate polarities. In reality, most laptops now have wire antennas that are run up the side of the LCD display and sometimes across the top. That gives it both a vertical and horizontal polarity. This is the exact same antenna design for your AM/FM car radio that is embedded inside the windshield.
Let’s start with the idea of how an antenna creates gain. Antenna gain effectively multiplies the signal being fed into it by borrowing the signal from other directions and refocusing it. For example, a 0dBi antenna is actually theoretical. It’s simply a point in space that radiates an equivalent amount of signal in every direction. Think of the center point of a ball. However, add a driven element and a reflector element, along a horizontal support arm at specific distances and you have a 2-element Yagi antenna that has 6dBi gain in one direction.
We will start with a 0dB antenna. A 0dB gain vertical antenna is really a 2.15dBi gain vertical antenna. That means it transmits 2.15db more power along the ground plane than it does straight up. If we make the antenna longer in multiples of the wavelength, then we get more gain. In reality, the highest non-collinear design I have seen is 12dBi in 2.4GHz. The resulting transmission pattern now gets squashed as less signal radiates upward and more signal gets transmitted along the ground for more range. Antenna theory is still developing with new algorithms coming out not only by engineers and scientists, but also by software programs that are discovering more efficient designs.
So how does 15dBi gain compare to 0dBi? In general, signal doubles in distance for every 6dB of gain. A 3dB signal gain increases the EIRP by a factor of 2. 6dBi gain would increase your EIRP output by 4 times which gives you about twice as much range. 15dBi antenna increases your range roughly by a factor of 6 times.
How does this play out in real life? Keep in mind that there will be obstructions in most areas. That means that getting ¾ of the way through a brick wall isn’t a whole lot more effective than getting ½ of the way through the wall. For walls or obstructions with less attenuation, we discussed how a 15dBi antenna can make penetration through an extra wall a reality due to a 6-8dB increase, or more, over the antennas that most metro APs used. A dual-polarity antenna with a lower gain can produce similar results. I have seen 2.4GHz multi-polarity antennas penetrate better than 900MHz single-polarity radios.
2x2 MIMO provides the option of 2 antennas, both in the same polarity or one horizontal and one vertical. There are even antennas that can do dual polarity or circular polarity in an omni-directional, or directional ,design. There are other variations on MIIMO, such as 2x3, 3x3, or more. If the antennas are directional, polarity is simple and cheap. If the antennas are omni-directional, vertical polarity is still cheap. Horizontally polarized antennas were much more expensive as gain goes up, but recent product releases demonstrate multi-feed dual-polarity antennas have come down significantly. Even dual-polarity parabolic dishes have dropped in price. We will cover these in future articles.
One of the more common horizontally-polarized antennas is the waveguide antenna. In an omni-directional design, they can deliver 13-15dBi, or more, of gain. Directional versions range from 14-18dBi. We used directional wave-guide antennas in some of our installations, and they worked great with 802.11b. One test we did demonstrated a laptop with a Cisco PCMCIA card connecting at 1.2 miles inside a fast-food restaurant.
Another popular design is the circular polarity antenna. The advantage to this antenna is that it transmits in all polarities simultaneously. The disadvantage over a single polarity antenna is that it sacrifices 3dB of gain for that multi-polarity coverage. Most circular polarity antennas are directional, although there are variations such as the Lindenblad design which is omni-directional. All antennas are compromises in terms of gain, direction, design, and cost. That is the reason it’s important to first define the target client before even considering any design idea.
There are a lot of variables to proper design of a system. Although the AP should be the easiest part, not including the antennas, beyond design scope, even firmware of the devices is important. Some APs handle packets differently than others. We discussed CPU overhead and real throughput in earlier articles. Firmware bugs and features also make a huge difference. Now throw in an antenna designs, network management, authentication, security, terrain, building construction, aesthetics, and even unknown challenges that occur after deployment, and this is when having a consultant who simply has more experience, provides value. However, even consultants are another variable, as evidenced by many differences of opinions and designs that have been deployed all over the world. Look for systems that are deployed and functioning and apply those ideas to your needs. Next month we get back to work since Grandma and Grandpa have now discovered Netflix.

Chapter 6 – Free is not a Business Plan

Our system is installed and our credit card maxed out. Now, we have to either pay for it or figure out how it’s going to save what we invested in it. As an income based system, it’s pretty easy to figure out a direct correlation between expenses and revenue. If there is some kind of defined savings, we need to try and make that objective and measurable.

Let’s talk about the profit scenario. These are just the direct costs:

1. 
   
   We spent $10K putting the system in.
2. 
   
   50Mbps costs $450 per month (data center plus roof rights)
3. 
   
   Pole rental costs $5 per month per pole (16 poles) or $80 per month

On the income side, you are going to have daily, weekly, and monthly clients. Let’s say you charge $5 per day, $15 per week, and $30 per month. It’s fairly easy to calculate your income/revenue to put a profitable scenario together. However, let’s go back to the original premise of a low cost system.

A municipal WiFi system has the basic problem of reduced range due to simply physics limitations. I plan to share additional ideas along this area in the near future but for now, let’s assume all clients are 2.4GHz and we still need 802.11G compatibility. This means that we either spend the money on an expensive, all-encompassing infrastructure, ala the sixty AP 2x2 MIMO design, or put that cost on the client side. Having the clients cover part of the Capex not only means a lower initial investment, but costs can scale upward with income.

This design took the original Muni AP concept, added 6dBi or better on the antenna gain, and had the benefit of 802.11N improvements in receiver sensitivity that adds another 10dBm. It doesn’t take advantage of 2x2 MIMO so we left 3-6dBi on the table of signal quality and bandwidth. However, we spent $10K instead of $100,000-$150,000. For 10% or less of the cost, we got 50% of 2x2 MIMO performance and 120% of the performance of legacy 802.11b/g systems. Don’t worry, there is a lot of capacity still left on the table that we can add later.

We now have to deal with the problem of not being able to connect to 60% of the indoor clients. This isn’t unique as most of the Muni-Wireless systems recommended some type of high-power indoor repeater device. Unfortunately, it was an afterthought when they determined that a high percentage of users couldn’t connect or basically that the system was grossly oversold. The indoor repeater balanced the power equation between high-power AP’s and weak laptop transmitter. The problem with these devices is that they create more interference on the channel due to that combination of high-power and omni-directional signal pattern. A better solution for the network is a directional client radio with higher gain antenna and lower power. There are many products but I suggest Ubiquiti Nanostation 2M or Nanostation 2M Loco radios. They have an optional window mount for indoor coverage and cost less than $100. They are also dual-polarity 2x2 MIMO in case the network gets updated later (hint, hint). The radios may need to be mounted outdoor for longer range or to get over the tops of houses or trees which means truck roll. These devices are not repeaters all you get is Cat-5 to the computer. Indoor wireless coverage will require a separate indoor wireless router .

How does this affect our profitability? Assuming 200 potential clients in 1 mile area, we need to get 18 clients at $30 per month to break even on the direct bandwidth costs, not including the payback on our Capex. That’s less than 10% of the potential clients in our 1 mile area, assuming all residential housing. Not an unreasonable number. There won’t be a lot of profit on residential truck rolls but at $200 per install, at least it won’t be a loss.

With 50Mbps per square mile and 70 clients, the system can be cost competitive with most wire line services. What happens however, if there isn’t a data center down the street? We have to figure out how to backhaul from a data center much farther away and probably within a LOS shot for a direct wireless. That could cost anywhere from $500 to $15,000 depending on distance, interference, and frequency availability on the roof. Although you could contact the local loop carrier and ask for a quote on bandwidth, the reality is you will pay $300-$3000 for 1.5Mbps to a 45Mbps DS-3 circuit. Some areas have MPLS and other data options but if you can get 10Mbps for less than $1000 per month from a local carrier, you are doing well.

Another option is to look for wholesale carriers for DSL. Although DSL usually ranges from 512Kbps to 7Mbps average, this goes up or down in an area based on distance to the Central Office or DSL switch. Assuming you can get 7Mbps down and 1Mbps up and your DSL wholesale carriers allows you to resell the bandwidth, you will probably spend about $60. Order 7 of them, put a Peplink 710 router on your network and you have 49Mbps down and 7Mbps up of available bandwidth. No individual gets more than 7Mbps down and 1Mbps up, but the router will load balance the users to get them the best bandwidth available. You are still below your $450 per month budget but the router will cost $4000. Peplink and other companies have smaller routers for fewer DSL lines starting at $300, so you can budget based on expected system needs. Keep in mind your oversell rate of about between 10-1 and 20-1 and that means 70-140 clients getting close to full bandwidth 100% of the time. 70 clients would generate about $2100 per month in revenue compared to your direct costs of $030 per month. The DSL idea can scale starting from 1 circuit keeping monthly costs in line with revenue.

The previous scenario is basically worst case. Assuming you have apartment complexes in the area, not only does the revenue potential increase, so does the percentage of temporary users. These are users that need 1 day, 1 week, etc… The revenue per day for 1 day users is 5 times higher than monthly users. Anything you can do to attract those users is a huge increase in revenue. Throw in areas that include business users, and the revenue potential goes up even further. Business users can be charged 40% more than residential users so there is more potential there also. Hot-Spots like restaurants, parks, etc… will add more revenue.

Here is where we are going to diverge from the original concept of mesh systems and open up the opportunity to make significantly more revenue. It’s been mentioned that the only way to really guarantee 100% performance of a mesh network is to install 60 AP’s per square mile. The reality is that it’s extremely difficult to recoup the kind of capital expenditure at $2500 to $3000 per installed AP (parts, labor, back end, and other miscellaneous costs) you need for this coverage and the monthly costs. Even our design, scaled out to its maximum potential down the road, will cost $1400 per AP installed (but it will it move some serious bandwidth). If it was easy to make a profit, companies would be throwing up municipal systems so fast; it would make your head spin. Throw in monthly costs of pole rental, backhaul or local loop costs, support, business expenses, etc…, and this model fails unless you get the following:

1. 
   
   The local government pays for use of the network thus supplementing the cost or by being the anchor tenant.
2. 
   
   Sprint, AT&T, Verizon, or some other carrier pays to hand off some of their subscriber bandwidth needs since their purse strings are slightly deeper than most of ours
3. 
   
   You find the 1% area in the country where wired carriers use their monopoly’s to make it easy to compete, there are lots of free vertical assets, and there are very few trees.

The system we designed achieves the strategy of 100% street coverage which meets most of the needs of public safety and municipalities. This opens up the government market. We have determined that some users will need indoor subscriber units. However, the one area that hasn’t been covered directly is the idea of the system simultaneously being used as a Point-To-Multipoint (PTMP) system. Basically we need a hybrid muni system. A PTMP system has the advantage of range but doesn’t provide street level coverage and usually won’t cover indoor. With an outdoor antenna on the client side, the system can support clients up to 2 miles away LOS. Our upgraded system will support up to 5 miles or more. This greatly multiplies the potential revenue of the system. Clients purchasing indoor units are creating a mini-PTMP system already. The only difference is that as the provider, you will have to provide staff that can go on-site and install a radio in a residential location. On the positive side, it can also be another source of revenue since the cost of equipment will be less than $110 for the install. Keep in mind that every subscriber we add brings in another $360 per year or more. This design with that addition, keeps the best of both worlds.

The focus of municipal networks has historically been high-density areas. The obvious advantage is having a market potential of 10,000 clients or more. These are the kind of numbers that are needed to cover a multimillion Capex. The budget model we created allows for much lower density deployment while still creating a design that creates a product that has value for public safety, water meter, parking, video surveillance, and other options that create value for a municipality to become a client. That provides two potential markets. Throw in the PTMP market, and we have not only created 3 markets, we can provide a more reliable, stable product with higher bandwidth capacity per client, and a larger coverage area.

Does this change the model of a true municipal network? Not really. Besides cellular, the most profitable wireless networks are PTMP. They cover many of the areas that wired never moved into. For example, I have an area of 50 homes that was never profitable for wired due to the length of runs. Put up a single AP with an omni-directional antenna, feed it with a T-1, charge $60 per month, and everybody wins. It beats satellite hands down and people still watch Netflix. Some part of the municipal network usually has roof rights for backhaul, usually on unlicensed frequency, to the AP’s on streetlights if they aren’t attached to fiber. Those locations are providing a PTMP system already for the AP’s. I’m suggesting that this same model be used for clients. In later articles, I will show you how this part of the network can be upgraded to easily deliver 20Mbps to residential and 50Mbps or more to businesses.

I’ve taken some heat for the fact that this system isn’t 2x2 MIMO. Keep in mind that this was first designed to create an inexpensive and/or profitable network. It will perform better than an 802.11b/g system due to better receivers, higher antenna gain, and better protocol. It can also support a PTMP design that can cover a couple extra miles around the 1 mile area for additional customers. The network is better controlled with more users using directional antennas for indoor coverage which reduces interference in improves s/n ratio. It doesn’t have 100% indoor ubiquitous coverage but it also doesn’t cost $150,000 per square mile, although it can be upgraded. We will next cover how to increase the bandwidth at each AP up to 80Mbps or so and expand the total capacity.

Chapter 5 - Reality, what a Concept

In the last article, AP’s got hung, WDS links were set up for hopping, and we were ready to attach to Internet. Although TriadLand is ready to rock, we now need to connect the users. First we need to attach connect the network to some type of Internet service and then come up with a way to authenticate the users that want to use it. After that we will cover the details on some of the system management and how to overcome them.

WISP operators are typically forced to work with local bandwidth providers who typically have some type of monopoly for the area. A WISP can’t resell most of the business services over DSL or Cable due to the local provider not allowing that. WISPs are generally resigned to T-1 circuits or more expensive business options. Assuming that you order a T-1 for your system, you now have 1.5Mbps of bandwidth for your network. If you plan on using cable or DSL services, check with your local provider to see if that is allowed. Other bandwidth options are also available but you will have to check for each area.

Assume that one of the 4 center APs out of 16 are the Internet connection point. We will set up the WDS links so that now end point is no more than 4 hops (meaning we may have to skip one) which will keep the last AP with around 5Mbps at the end point. If we can skip more than one AP, we can keep the hops to 3 or even 2 if we have LOS between the AP’s. We have enough signal to hold very high modulation rates with ½ mile links between APs. Keep in mind that APs between the end point and the egress point will be handling users while simultaneously passing WDS backhaul for other APs down the chain. This will directly affect throughput for users down the chain. It is one of the limits of this design but no different than any of the other earlier mesh designs. The end result is that the entire square mile will eventually be routed through one AP.

One key issue I received a couple of emails on involved security. The drawback of inexpensive radios is that you may have to give up something in return for the reduced price point. This system provides no encryption over the WDS links. This problem gets resolved with additional hardware as part of an upgraded system. The system can run security between the laptop and the AP but there isn’t much use if the AP hops aren’t secure. If you plan on upgrading later with more bandwidth, then it might be a good idea to get the users to use WPA2 on the APs from day one so they won’t be confused later. Just make sure your EULA clearly states that the system is not secured over the wireless link.

The second problem with this network is that it only supports a single SSID. I do not know if that is going to change in the future. There is third party firmware that will run on the Bullets that may offer more options but that typically comes with additional costs which gets away from the original premise. If you need security, then VPN tunnels are the only option with the basic system. Phase 2 resolves most of the security issues.

There are many good products out on the market for authentication of users. Our sites use Patronsoft Firstspot for user authentication and management. FirstSpot runs on Microsoft Windows XP or Windows Server, can support SQL Server for extending site deployment and centralized user management, uses PHP for the web pages, and can run fail-over servers for offline management. Since our company has years of experience with Windows, it works for us. Those of you with more experience with Linux have many other options. We ran tens of thousands of users through our servers over 5 years so I’m pretty comfortable with it. However, it’s not CALEA compliant yet but they are looking at it now.

Triadland is up and running. Users attach to the broadcast SSID, get a login page, diligently read the EULA word for word in which they agree to follow the rules, and then they get online. Now what happens? This is stuff normally planned out in advance but the article focus was determining the basic wireless technology first. It’s time to deal with the actual functionality of operating the system.

This is where every WISP’s worst nightmares start. It begins with the Federal government getting access to your system in the name of Homeland Security and ends with junior making it his personal mission in life to download the entire Sony movie collection. Your first job is to file your CALEA paperwork. CALEA is an entire article by itself and I may cover it much later. Go to http://www.wispa.org/?page_id=2022 for more information. After that, you need to figure out how you are going to keep control of junior. You are also going to have to deal with the users that move large amounts of spamware or spyware without even knowing it. These same users can get you disconnected from your Internet circuit if it’s bad enough.

Let’s move on to the first issue which is keeping control of your network. Several things will stress both you and your network. Let’s start with junior’s desire to fill up that new 2 Terabyte hard drive he just got for his birthday. File-sharing is one of the biggest problems faced by most ISP’s. Fortunately or unfortunately, depending on which side of the equation you are on, recent rulings by the FCC allow operators to limit file-sharing. There are 2 basic ways to handle this. You can use either an authentication server that keeps track of bandwidth used or a web application firewall that allows you to block file-sharing applications. Limiting users to 10GB – 20GB per month is a good start.

Early on, we had an incident where some users on our network had been infected and turned into spam servers. The bandwidth provider started blocking Internet for the entire system until we got it resolved. With 200 plus users and some of them still running Windows 98, the battle to keep viruses and spam under control is difficult at best. We purchased a Barracuda Web Application server which not only blocked the offending users, it redirects them to run spam removal software and won’t let them on Internet until the computer is cleaned. A Web Application filter also allows blocking of specific websites that might cause unwanted legal attention and file-sharing applications.

Now that we have built our 1 square mile network in Triadland, our next step is to make either make it profitable or find a way to give it purpose. We will cover those ideas next article.

Chapter 4 - You Should Never Hang APs With Your Spouse

The low-budget Muni-wireless system is ready to be deployed. We have identified the perfect one square mile area to be covered. The streetlights are 25’ tall and exactly 660 feet apart with full time electric power. There are no trees and the houses are 20’ tall and made of wood. Nobody in the area owns a microwave oven or any indoor WiFi routers. This will be known as TriadLand. It’s my make believe city. I get to name it. The goal for the first square mile is complete indoor coverage to a laptop. Let’s find out if that is realistic.
First calculate the ideal distance that a laptop can connect to an AP. To do that, get the technical information on the AP, the antenna, and the WiFi card in the laptop. The AP we are going to use for this design is a product manufactured by Ubiquiti, the Bullet M2 HP (http://www.ubnt.com/bulletm). The antenna is a 15dBi omni-directional (tested closer to 14dBi) collinear vertical polarity design sold by L-Com (http://www.l-com.com/productfamily.aspx?id=6414). Laptop specifications will have to be an educated guess, since they will vary. We will use 15dBm for the power output and an antenna gain of 1dBi for all calculations. I’m going to assume the receive sensitivity is the same as the AP. Although the Bullet 2M HP is certified for only a 6dBi antenna, I have already addressed the issue and this combination will pass FCC rules in a few weeks.
Astute readers will notice that the firmware included with this radio doesn’t support mesh. However, it does support WDS. Basically, each AP will have to know the Mac address of the AP before and after it. This had to be entered manually into each AP. WDS then creates a layer 2 connection between each AP. By using WDS instead of mesh, we lose 2 things. The first is that if a radio fails, that path fails. There is no recovery option other than to drive out and replace the AP. However, it’s also possible to connect to the next AP wirelessly and manually bypass it. The second is that the system does not provide any type of load balancing. Not every mesh system does so that anyway. We also have an upgrade path if there are clusters of high-bandwidth users that we will cover later.
The Ingress/Egress point has to be located so as to minimize the number of hops. We will have a useable 50Mbps TCP/IP drop to about 10Mbps at 4 hops (I know that doesn’t calculate right but it’s processor limited, not bandwidth limited). I will cover an upgrade later that delivers an additional hop and expands the first hop to 100Mbps or more. Keep in mind UDP is much higher.
Mounting this AP/antenna combination requires some ingenuity in terms of bracketing and power. I am working in a simplified mount for it now. The antenna is provided with a simple U-Bolt for up to a 2.0” vertical pole. The radio simply screws into the bottom of it. Power requires 12-24dc volts. Interestingly enough, the power issue is nice for a solar application. However, it requires a PoE that although less than $10, is not designed for outdoor mounting and is 110V. If the radios will be mounted on horizontal poles, the system will require additional bracketing for this installation. The power problem will require a small Nema enclosure to seal the AC/DC converter along with a Metropole power adapter or an electrician can mount a power outlet inside the pole.
A very simple formula for calculating the link path budget is (Power Output of the AP) + (Antenna Gain in dBi) + (Receiver Sensitivity at the speed we need). The signal loss equation tells us what we are going to lose in signal level over distance or Signal Loss = (20 x Log10 (Frequency in MHz)) +( 20xLog10 (Distance in miles)) +36.6. Since all of the connections are very short range, LOS or NLOS, and well below 10GHz, we can sort of ignore a lot of other variables that are used for longer range and higher frequency calculations. So how far can we get 2 AP’s to talk to each other and keep a maximum modulation rate with a power level that matches the laptops?
Since the EIRP of the laptops is about 15dBm, we will start there. The link path budget between APs is -
802.11N: 15dBm (radio output) +28dBi (antenna gain is a summary of both sides) + 74dBm (speed at MCS7, 65Mbps, for the Bullet M2 HP) = 117dB
The link path budget for legacy laptops is -
802.11b/g: 15dBm (radio output) +15dBi (antenna gain is a summary of both sides, assumes 1dBi on the laptop) + 74dBm (speed at 54Mbps for the Bullet M2 HP) = 104dB
The second equation needed is the free space loss. I’m not going to go into the formulas here, but the result calculated at 660 feet was -83dB for the laptops. That means that we have an expected signal level of -56dBm. From AP to AP, we should see a signal level of -49dBm. So in TriadLand, we could install APs at 1320 feet with signal to spare at the full 802.11N or 802.11g bandwidth while delivering the same signal level on both sides.
The reality with an urban deployment is that the farther down the street the AP is located, the more houses the signal has to pass through. We will assume that the AP is mounted on a street light. Houses built of brick, stucco, or aluminum siding along with trees, bushes, and Uncle Bob’s motor home in the driveway, all add attenuation to the signal quality. Therefore, for the sake of this design, we will assume that each house adds about 10dBm of attenuation to the signal. The Bullet 2M, at the lowest connection speed in 802.11N mode, has a minimum sensitivity level of -96dBm. A good rule of thumb for noise in 2.4GHz is about -85dBm. That means we have -29dBm of attenuation room to make a connection or about 3 houses. However, since we want a 10dB difference between signal and noise, we now are down to penetrating 2 houses at maximum speed and dropping off from there.
Attenuation is the biggest problem in an urban setting which means that LOS is more important than distance. To guarantee connectivity in the worst locations with trees and brick houses, it will take APs about every 1/8th of a mile which means no client is further than 330 feet. Cities such as Mountain View, St. Cloud, Scottsdale, and others learned this the hard way after setting up the system with fewer APs per square mile than really needed. In addition, those APs used 7-9dBi. We are adding up to 6-8 more dB gain on the antenna side which should add about another house of penetration.
Being more realistic, what are the options? Let’s say that we assume 16 APs per square mile and all clients are 802.11N. Also assume 14 houses per block and 2 blocks for 1/4 mile. The light poles will be about 30’ from the front of the house. The farthest house is 660’ away and will not have LOS except for the first couple of feet into the front of the house. In the worst case, users may have to penetrate 8 or more houses at various angles, plus trees or vehicles may be in the way. If the houses are brick, stucco, or have aluminum siding, the signal will definitely not go through it. In this environment, we are installing 25-49 APs. The total cost for a 16 AP install is about $10k per square mile. As mentioned before, $1000 upgrades the bandwidth to 60Mbps for the square mile and 10Mbps at 4 hops. If you need more bandwidth than that, the cost jumps to $14K per square mile per 16 APs and carries a 100Mbps pipe through every AP.
Based on this analysis, you now have to make the decision as to the purpose of the network. If it’s ubiquitous coverage, then you have no options. It’s going to be a minimum of 25 APs up to 49 APs or even more in the worst environments. Even though the cost of each radio and antenna is $200, the cost of installation will drive this to about $14K-$42K per square mile with that many APs. That sort of defeats the idea of a budget system. If you were planning on this type of budget, then there are additional changes that should be done for many other reasons that we will cover in future articles.
If we just wanted outdoor coverage only to laptops on the main streets, 16 APs are probably more than sufficient. If the goal is simply street level coverage for cameras on light poles and vehicles, 2-4 APs may cover that. By eliminating the 100% coverage for indoor 802.11g laptops with their limited power and antennas, the system has more options and range goes up significantly. There are so many variables in a metropolitan deployment due to the physics of microwave frequencies, that $3000 per square mile in Arizona might become $18,000 per square mile in California.
Now that we know the realities of RF, let’s stay with the original concept and change the expectations of the system to meet our budget.
1)100% street level coverage for cameras and public safety vehicles
2)100% street level coverage for laptops and portable WiFi devices (this does not mean back yard and between houses)
3)20-40% of the residential locations will have 100% coverage
4)20-40% of the residential locations will have varying coverage within the house
5)20-40% will require indoor equipment or a professional outdoor installation of a client device.
6)The system will support legacy 802.11g equipment
7)Minimum residential bandwidth from 1Mbps to 5Mbps
8)20 Mbps per square mile total available bandwidth (upgradeable to 60Mbps per square mile for an additional $1000)
9)Expandable up to 300Mbps per square mile (for an additional $4000 per 16 APs)
Our second problem is user density. Given that most APs are limited to 20-30 users per AP and each AP covers up to 80 homes, this isn’t an issue. However, in some cities with apartments, density could reach into the 10,000 people per square mile range. That could mean hundreds of people per AP. If you find a place like that, don’t worry, we have plan C.
We now have to go back to the noise issue we discussed earlier. Typically you want the signal to be at least 10dB better than your noise. In Triadland, our noise figure is about -96dBm. In the real world, it’s going to range from -65dBm (bad) to -85dBm (reasonably good). Since our signal level at 1320 feet between APs with a 15dBm output is going to be -50dBm, we should be able to handle even higher noise areas. Laptops should be connecting LOS at about -62dBm at 660 feet.
In some areas of the country, and this is the part I like, the scaled system could deliver 20Mbps - 50Mbps to a residential or business location. Based on some numbers I received a few days ago, Internet bandwidth can be purchased for as little as $1 per 1Mbps per month at local data centers. If a residential 10Mbps rate is sold for $30 and the over-subscription rate is 10-1, the potential revenue on a 1Gbps circuit is $30,000 month assuming you have a way to transport it to your WiFi area. There are obviously a lot of capital costs here. This will also take serious engineering and planning involved for this type of revenue stream, but it’s now possible. 4 years ago, nobody was making a profit at this. IMHO, this now means that wireless can directly compete with cable and DSL services.
This design also solves another problem that WISPs are running into. If you read the last article, it’s evident that WISPs using unlicensed frequencies are running out of bandwidth. This design shortens up the distance from the client radio to the AP. Instead of connecting at a few miles and increasing the risk of interference, all clients are within a few hundred feet of an AP; thus, minimizing the interference issue. It also lowers the required altitude path.
Our basic network is defined. The budget is figured out. Our users in the first square mile of TriadLand have connectivity to the network. Now, how do we get them connected to the Internet and keep track of them? We will cover that in the next article.

Chapter 3 - Share and Share Alike

Although this was the week that we were going to discuss putting APs on the poles, an incident occurred last week that I think is worth discussing before we go any farther. Always keep in mind that unlicensed bandwidth is a shared commodity. It’s also a good idea to have a relationship with your competitors or anyone in the area that is using unlicensed bandwidth. Sometimes that’s a little hard to do, but it’s definitely in your best interest to try.

Traffic lights and cameras are a perfect application for 4.9GHz or 5.8GHz radios where fiber isn’t installed. For example, we connected several traffic lights in Glendale, AZ a month before the city hosted the 2007 SuperBowl. The entire network consisted of 5.8GHz SkyPilot equipment with Pelco cameras. The system is used to monitor the intersections and also as the SCADA system for the traffic lights. This allows the city to manually adjust the traffic lights before and after events to optimize traffic flow. The city just recently closed on a bid that ADOT managed for the next phase. For some reason that I still can’t explain, they removed the compatibility clause with the existing system and simply went for low bid. Due to the cost of the SkyPilot equipment, it was obvious that low bid would not be SkyPilot or a compatible system but something else. This means the city will now be supporting multiple mesh networks along with the extra long term costs. He who controls the funding; writes the rules but apparently doesn’t have to support it.

That led to another project to connect 13 traffic lights with 50 High-Definition cameras at resolutions up to 1600x1200 pixels. We settled on Axis cameras for this project. The ideal cameras, fixed and PTZ, would be 1080p, 24 frames per second, and H.264. The project came close with most of these specs, but there were some compromises. We either got PTZ and H.264 at 720p, or we got 1600x1200 at reduced frame rates with MPEG-4. Future intersections may get 1080p cameras. This was sufficient to meeting the system requirements for the application which was traffic light management, license plate recognition, video analytics, and future facial recognition software.

Either way, we needed a lot of bandwidth at the City Hall to make this work. There were no other antennas on City Hall. A quick site survey showed additional equipment in the area with fairly low signal levels. The AP chosen was Ubiquiti 802.11N Rockets with 90 degree, 20dBi dual-polarity sector antennas. The system has 4 APs and sector antennas on City Hall for 360 degree coverage. Each light depending on distance to City Hall gets either a Nanostation 5M (integrated antenna) or Rocket 5M (external flat panel sector antenna). We are also adding 2.4GHz 802.11N to all the intersections for future use. It only adds $200 to the cost of each intersection and has an 800’ range to a laptop.

The first AP was turned on about 6 months ago and covered one 3-way intersection with 3 cameras and a second client station for data backhaul to one of the City Parks. The park had maintenance staff and a camera for the skate area. The second AP was scheduled to be turned on this week to cover the next 90 degrees off City Hall. That was when we discovered that between the time we turned on the first base AP and then went back to turn on the second base AP a couple of months later, the IT department had a local WISP put a Motorola Canopy radio on the roof without notifying the traffic department. Even more fun, it was in the 5.8GHz band.

Our traffic system was designed to use all 100MHz of the 5.8GHz band and was designed with signal levels for each link between -50 and -65dBm. Since the general rule is at least 10dB headroom on the link path and the minimum signal for APs with an MCS15 link (130-144Mbps modulation rate), 2x2 MIMO is -75dBm, you try to design for your worst signal at the receiver to be -65dBm. At those signal levels with directional antennas on the client side at 2 miles or less, it’s fairly hard to interfere with this system.

As a courtesy to the vendor, I thought that giving them a call and asking if they could change the Canopy to 5.1-5.4GHz before I turned on any more APs would be the professional thing to do. Unfortunately, when you run into inexperienced WISP technical support staff who think that whatever product they use is significantly better than whatever product you use, there is a problem. The conversation started with me asking if there was any chance that they had other frequency options. The technician asked me what equipment I was using. I replied. He then made the comment, “Our equipment will stomp all over your equipment.” I quickly established that this wasn’t the person to which I needed to speak to resolve this issue.

Every wireless product handles interference in different ways. Better filtering, multiple polarity, diversity, beam-forming, better firmware and many other techniques are utilized to improve the quality of a connection. Obviously some equipment works better than others in high-interference environments. However, 802.11n doesn’t always play very well with 802.11 b/g or 802.11a radios. In addition, physics are physics. Throw in a 2x2 MIMO signal with very a high gain antenna and a total EIRP of 42dBi minimum on both polarities, and there is going to be interference. This is especially true if the competitor’s base station is 1000’ away and their system is based on a 30dBm EIRP signal level from the APs. Since the traffic system was designed with a very high level signal path with multi-polarity 2x2 MIMO, there wasn’t much concern that the local WISP was going to interfere with the traffic system. On the other hand, it was clear that the traffic system was going to cause some interference to the WISP operation. I was wondering if that was already the case.

The last thing I wanted to do was start a war between a city government and a local WISP. Since the WISP was already serving some of the registered voters, this was probably not a good idea. The ego in me wanted to turn the remaining two sector antennas directly at the WISP base station across the street, program the two radios with a channel width of 40MHz to cover most of the 5.8GHz band, turn on 802.11N mode only, start multicasting “Transformers 2” in UDP mode to my other laptop and phone, and crank the power to 48dBm EIRP. I felt there might be a lesson to teach on RF, signal-to-noise ratio, and how to be polite when other companies are trying to do the right thing. Instead, I called back and talked to upper level management to arrange a meeting to see what can be done. That’s when I found out as I had suspected, that the traffic system’s first AP possibly might already been causing some interference issues. Unfortunately that traffic system AP also serves the City Park and has been running for several months. As a courtesy, the best I could do was to turn the power down until we plan out a more cooperative strategy with the WISP.

Now we can come back to our design. The AP combo we are considering will have an EIRP of up to 36dBm and will be 802.11 b/g/n compatible in 2.4GHz. That is far more powerful than most indoor systems. There are ways to limit the interference and everyone sharing those bands should consider this. It’s good policy to always engineer for the least amount of power that you need. Also, you should consider using the least amount of bandwidth. If you don’t need a 40MHz channel, don’t use it. You gain 3dB on the receiver side with no more power if you use a 20MHz channel. Drop to 10MHz and you pick up 3dB more, etc… The tighter the channel, the better chance you have of getting a signal through.

There are many of you that may not even want b/g compatibility. If the system is only used where you control both the APs and the client radios, then you have some options. In that case, using 802.11N only is the best way to go. This results in fewer APs and better performance. It’s also possible in this same scenario to get as many as 100 users on an AP or as many as 300-400 on a single pole. If all the equipment is the same manufacturer, consider a 5-10MHz channel which will still deliver 12-25Mbps per AP. Running 2x2 MIMO doubles the throughput to 25-50Mbps in a 5MHz channel. This is probably more than most of us need for a link. With a 5MHz channel, there are 6 channels to work with in 2.4GHz instead of just 3.

Whatever gets installed will probably interfere with other existing systems. We have to be cognizant of the fact that nearby systems might be critical for hospitals, voice systems, video surveillance, office networks, etc. For example, a few years ago, we saw an outdoor AP in Scottsdale stadium take down a wireless voice system at a nearby hospital. Since we put our phone number in the SSID, they could contact us quickly. We adjusted our antennas and resolved the issue. Site surveys have to be done everywhere to understand the ramifications of the install in addition to understanding your own coverage zones.

Based on all this, if I saw a little Motorola Canopy radio on the roof and thought it was a good idea to possibly contact that vendor, what was the installer thinking when he saw all the Ubiquiti equipment along with a Motorola PTP600 radio? He obviously didn’t do any kind of site survey. The reality is that this is standard procedure for most companies.

Many companies that rent roof space for APs do it from some type of property management company. There is usually no bandwidth management on the roof. When someone finally overlaps frequencies, an avalanche of problems will start. For example, administrators may turn up power to overcome the interference if they have no other options which causes even more problems (this gets into noise and channel filtering). The responsible thing to do would be to notify all parties involved and see if there is a way to work together. Obviously this sentiment is not shared by everyone, or some field installers aren’t trained well enough to watch for this situation. Then again, they just may not have cared.

Unless I run into another tech support person who hasn’t been taught the concept of cooperation and I need to vent, we will get back on our project next article. I’m still working on some FCC equipment certification issues with the design. Not that we can’t do it but I want maximum performance and the equipment I need isn’t quite yet available. I’m also waiting for new products to test that may enhance the system. Anybody can build a system with unlimited budget. The fun is creating a carrier class dependable system on a shoestring budget; since, that’s what most cities have in their coffers these days. Any new products that have the potential to help preserve that coffer is going to get their attention. Next article, we get back to work.

Chapter 2 - Access Point Secrets You Didn’t Share with your Parents.

Continuing the discussion of the budget Muni-Wireless network requires analyzing the front line component - the Access Point (AP). There are many variations of APs. They all have features and capabilities that provide enhancements in certain environments. Some of the APs break the practical limit of 20-30 users by using multiple radios in a single enclosure, proprietary polling systems (not compatible with other vendors), and advanced beam-forming techniques. However, the focus is from a budget standpoint and that means this design will start with a single 2.4GHz radio with an omni-directional antenna. Later articles will cover upgrading the design to support more users and a larger coverage area. The beauty of this design is that it’s cheap to get in the game and scalable.

The ideal inexpensive AP was covered in the first article. This article will next cover what protocol it should support. Obviously if it supports 802.11n, that’s huge. 802.11n can provide up to twice the range of 802.11 b/g with the same single AP, single antenna design. This is due to a combination of a more efficient transmitting modulation scheme and better receiver design than 802.11b/g/n. In 802.11n, this would be considered a 1x1 MIMO. If the world was perfect and everybody had 802.11n devices, theoretically it would take only ½ to ¼ as many APs per square mile for the basic system. Unfortunately, the expansion of WiFi phones and their lack of support of 802.11n means keeping legacy 802.11b/g compatibility, range, and load in mind. If the network doesn’t need to support 802.11b/g the savings are huge. The reality is though, if the network was built today for the general public, the system should still support 802.11b/g equipment.

Taking the AP design a little further, figure out what the client connectivity area is. If the goal is to connect to Joe or Jill Teenager, who lives in suburbia with maple trees covering the sky and every home is built like the Windsor Castle (brick), while he/she is lounging on the couch with his/her iTouch which doesn’t support 802.11n; then even with 16dBi omni antennas, the AP will have to have to be within a couple hundred feet. If they live in Stuccoville, Arizona, (yes, stucco attenuates signal, but go with me on the flora thing) the range will be much greater since trees are 60’ tall with leaves starting at 55’ (palm trees). The only other vegetation is 2’ tall and is considered a lethal weapon (cacti for the Northerners). This is a slight exaggeration, but the basic physics stand. If the clients are outdoor only such as surveillance cameras, mobile hot spots for police, utilities, etc…, then the number of APs is reduced significantly. In fact in Arizona, there are places where 2 APs per square mile are all that’s needed for mobile coverage or every traffic light corner.

The really interesting thing concerning small deployments, parks, mobile home parks, etc…, is that the back end Internet bandwidth options are rarely as fast as the radio capacity for cost reasons, access, or End-User-License-Agreements. For example, one park that Triad manages is really only under heavy use for two months during the year (Spring Training) in addition to 2-4 high-use specialty events. Bandwidth options are a T-1 for a whopping 1.5Mbps ($450 monthly), DSL for 12Mbps ($100 monthly), or cable for 4Mbps ($200 monthly). Assuming 12Mbps is enough (it can be expanded with multiple DSL circuits as needed), users are limited to 2Mbps, the radios support up to 20Mbps in 802.11g, and there is only 2 hops maximum meaning no need for multi-radio APs. Even at 2 hops, bandwidth is still 10Mbps on the perimeter. There are 11 radios total to cover about ½ square mile with 2 high-density cluster areas. The 2 high-density areas are each comprised of 4 separate radios at the perimeter locations. The remaining areas are covered by 3 other radios. Using high-gain omni-directional antennas, laptops can connect at 800-1200 feet. All radios are in AP+WDS mode for relay and coverage. The total cost for all the radio equipment and antennas for this deployment was $2300, not counting the tower mounting brackets. The system has been up for over 2 years and is scheduled for a planned 802.11n upgrade in the next month or so. The omni-directional antennas will stay, meaning that the 2 hop system will now deliver about 20-30Mbps on the second hop.

The project utilizes four vertical assets up to 150’ that also include a 5.8GHz PTMP system. The 5.8GHz PTMP system has a range up to several miles in a 360 degree pattern. Each vertical asset consists of one radio with a standard vertical polarity sector antenna. When the PTMP system is upgraded, it will consist of 5.8GHz 802.11n, 2x2 MIMO equipment. The cost of the upgrade will be approximately $250 for the radio and antenna and will support up to 100Mbps per radio. Throw in another $100 per pole for backhaul and the system has now expanded from a simple hot-spot project to a 100 square mile 400Mbps infrastructure for an additional $1400.

It’s time to step back and discuss the unadvertised secrets of 802.11n since it affects the expectations of the design. Here’s a big shocker - the processor and firmware of the AP affects the radio performance. For example, the AP manufacturer claims 300Mbps. That’s modulation, not real-world throughput. That number is also rounded up from the real number which is between 270-288Mbps, depending on what’s called the guard scheme (not within the scope of this article and makes my head hurt). Keep in mind that many of the 802.11n radios also have 100Mbps Ethernet jacks because manufacturers know that communication is half-duplex. The Ethernet port is full-duplex so they consider 100Mbps up and 100Mbps down, 200Mbps total, not 200-300Mbps in one direction. Strike one.

The second issue is that 300Mbps is only achievable by running 40MHz wide channels. That works great in the living room for 50’. It doesn’t work so well when the AP is perched on a light pole with 200 houses in range. It would be difficult at best to go 500’ in 2.4GHz and get maximum theoretical modulation rates with a 40MHz wide channel. Multi-radio APs typically use the 5GHz bands for backhaul for that reason. That means the real-world useable 2.4GHz bandwidth is 20MHz which translates to 150Mbps. Strike two.

The 5.8GHz band is the most commonly used with 40MHz channels. 5.1GHz to 5.3GHz is usable with some manufacturers, but the EIRP drops significantly. However, for 500’ and no vegetation, that is reasonable. There should be very little interference in those bands. Of course, many of the radios that are already in those bands are probably set up by a local WISP. Although most WISPs are knowledgeable and legal, there are a few WISPs that either have no clue about the rules or are intentionally broadcasting illegally due to congestion in the 5.8GHz band. There are manufacturers who have certified equipment in those bands, but because of the limited EIRP in those bands, the equipment isn’t as popular. 5.8GHz also isn’t supported by most laptops and smartphones.

Assume that there isn’t interference in whatever band with a 40MHz wide channel. Then the next question is how far can a 300Mbps modulation level be maintained? Well, here is the second problem. To quote a good friend of mine, “speed, distance, reliability, cost - pick 3”. The first 2 are the most critical. For just the radio, it should be “speed, distance - pick one”. Basically, to get the 300Mbps modulation rate, receiver sensitivity drops to around -72- to -74dBm and the power output of the radio drops from the fabled 26-30dBm to around 24-26dBm or less for outdoor equipment and 15-18dBm for the retail $30-$200 equipment. To get the magical 300Mbps speed in a 360 degree coverage pattern from a single radio with an omni-directional antenna, there should be total LOS, no interference, and at maximum distances of around 500’ or less with the 40MHz channel. Most multi-radio AP manufacturers recommend directional antennas but one could argue that defeats the concept of ubiquitous mesh architecture. Strike three.

I have not done testing with most of the new 802.11n APs so some of you can jump in here with some real-world values. I’m also not going to get into sector antennas here for pole installations because the size of the antennas which makes it difficult to get through city zoning. I’m going to also get some grief here from the beam-forming guys; but realistically, none of the beam-forming systems in 2.4GHz can match a 28” tall sector antenna in performance. It’s a simple matter of physical capture area. However, newer sector antennas also support multi-polarity which gives them even more advantage.

Although baseball doesn’t have a strike four, here it comes. 300Mbps modulation rates are a real-world throughput of 150Mbps through one radio under absolutely ideal conditions. This typically doesn’t exist in most suburban realities. In addition, the processor is also a bottleneck for the AP. On some 802.11n devices, a straight ftp transfer from one computer to another computer with a 20MHz wide channel (much more realistic) will result in a transfer of about 35Mbps. Put two computers on each side, and then do the same transfer computer to computer, and the throughput jumps to about 60Mbps with each stream about 30Mbps. Do a 3x3 transfer, bandwidth goes up to 72Mbps, and each unit drops to around 24Mbps. Further testing up to 10x10 transfers demonstrates 85Mbps maximum. All of these numbers are perfect signals at short range in the lab. So what happened?

Basically, some processors/chipsets get more efficient as multiple transfers of data occur but are fairly limited for a single transfer. For example, using a 20MHz channel and MCS7 modulation rates (65-72Mbps depending on guard scheme), will result in a 60Mbps with 65% CPU overhead with UDP video traffic from a single camera. However, change that to TCP/IP and the rate drops to 35Mbps and CPU processor overhead jumps to 100%. This test was done with a 400MHz Atheros processor and chipset in the radio. Other manufacturers also use the 300Mbps but really add in the combined throughput of 2 radios, not a single data transfer rate. Keep in mind that this isn’t for every single product out there, but it shows that real-world testing is definitely required before a design is signed off on for the particular application.

802.11 a/b/g APs exhibit this same behavior when compared to the marketing material so it’s not a new phenomenon. In some cases, manufacturers used UDP traffic numbers instead of TCP/IP real-world traffic. In other cases, manufacturers were using 40MHz channels for their specifications which again, aren’t realistic for most installations. In many cases, when 40MHz channels were used, the CPUs capped out so that the result was not a doubling of the transfer rate but maybe 80% more over a 20MHz channel. For example, 20Mbps at 20MHz translated to 35Mbps at 40MHz.

Another problem with processor overhead is how many packets per second (PPS) that can be jammed through at one time. A user opening up a web-site usually has a fairly low PPS requirement, thus low CPU overhead. Open up a file-sharing application and the number of sockets and PPS can go through the roof. Low-cost APs trying to handle this kind of traffic typically slow down drastically. More expensive and faster processors along with better firmware scale better under load.

There are many reasons that one AP costs $100 and another costs $6100. More expensive APs will use 2, 3, 4 or sometimes more radios within a single AP enclosure. The rest of the costs come in the form of firmware, management tools, and capabilities. Many of the outdoor units have firmware designed to optimize video transmission quality, fast-handoff between AP radios while vehicles are moving, mesh implementation, beam-forming, multiple SSIDs, multiple frequency, channel bonding, and many other advanced WiFi technologies. 802.11n radios are also start with a better specification foundation.

Switching back to the budget system, WDS communication is built in to almost all the chipsets and supported in firmware. This allows APs to connect to each other through Layer 2. It’s simple, sometimes not compatible between manufacturers, and requires manual setup with MAC addresses of each device. Mesh firmware, with a basic setup, simply finds all other units in an area, then sets up either a layer 2 and/or a layer 3 network between the radios dynamically, as part of its underlying mapped infrastructure. Mesh firmware is always proprietary between APs.

The goal of Tales from the Towers is to create a budget, scalable, municipal WiFi system. It will probably never have many of the capabilities that more expensive systems can do. The budget WiFi system will not initially and probably may never support fast handoff for moving vehicles, optimized video traffic, or possibly multiple SSIDs. Therefore it’s important to understand what its capabilities are before deciding that price is the most important factor. In some cases, its inexpensive nature, flexibility, and scalability allow the network to get around many of the advanced features that some of the more expensive APs have. In others cases, it simply isn’t going to happen and the proper design should be Motorola, SkyPilot, Firetide, Tropos, MeshDynamic, Ruckus, BelAir, Meraki, Meru, or whatever product fits the needs and budgets of the municipality. Every one of these companies has features and benefits that make them ideal for specific applications. I have designed and installed systems using many of these products.

However, the focus here is a starter system the meets the cost and scalability of really tight budgets. This makes it deployable in areas that can’t justify the ROI of more expensive equipment and can bring Internet to areas that may not be able to justify a large investment. The next article will cover the actual AP deployment design.

"Rory Conaway"