My attention deficit disorder is going nuts so I’m going to veer off Triadland a little bit more. We have defined a reference low-cost system and we clearly discussed the fact it’s definitely not the best system for a public safety mobile environment. So I’m going to call this article the prequel. How do we determine what equipment to use or in other words, what is the best system? Actually, most of them have some unique value depending on the application.
Ok, that’s a copout and allows me to avoid the question. However, there is no easy answer. I’ll go through the process that I use when asked to design a system. Keep in mind there are way more variables than I can cover here. The other thing that is highly important is that you know as many products on the market. I’m not just talking about mesh products, but PTP products, PTMP products, mobile products, virtual IP software, VPN products, antenna designs, etc… In addition to reading and keeping data sheets from everybody, get your hands on and get as much field experience with as many manufacturers as possible.
Almost every design I create involves products from multiple manufacturers. Most designs have some type of budget restraint but every design has some function that has higher priority than others. In some cases, the project was referred to me by a supplier or manufacturer. When that happens, I have a personal obligation to work with their particular product lines. That hasn’t been an issue so far but sometimes it means reaching deep into the product toolbox to find other products that enhance a design. Being vendor agnostic is very important for a consultant.
For example, one manufacturer came to me to help design a PTMP system using their APs. The problem was that they didn’t really have an outdoor product line available yet. They were working on one but in the meantime, they designed a NEMA enclosure and cabling structure that allowed the indoor units to be used outdoors. The basic problem with most indoor units was that the power output on the transmitter is usually underpowered compared to most outdoor units. FCC certifications with the manufacturer antennas were also clearly going to limit the range of the equipment.
In this case, I first had to do a little legwork to understand the rules I was limited by. I then had to analyze in detail, the bandwidth requirements at each location to determine what PTP or PTMP design options I had and what modulation rates I was going to be able to achieve on each link. This was to ensure that I didn’t overload the links. Since this design was going to start with 5.8GHz links, I was going to be limited to MCS rates up to the magical 300Mbps depending on distance. As we discussed in earlier articles MCS(15) rates are hard to get but in reality, I’m not getting more than 80Mbps TCP/IP in one direction. Even that was going to be difficult. The other problem is I didn’t know the APs in terms of actual throughput with video streams so that’s another variable where I have to err on the side of caution. It’s also probably a good idea to know the chipsets the AP manufacturers are using and the FCC AP filing information to review manufacturer data to get the actual facts that aren’t in the brochure.
Given the distances, FCC regulations, bandwidth requirements, and a few other unknowns, it was evident there was no way this equipment was going to be able to handle all the traffic. That’s when the design changes and you have to go to different manufacturers for at least some of the project. The important thing to me is to still use as much as the referring supplier or manufacturers equipment as possible.
For example, let’s say that we have a single primary building connecting to 2 other buildings that are within 30 degrees of each other from the primary building. We will call them A, B, and C. The A building is the primary, the B building is 1 mile away and the C building is 2 miles away.
I have 3 options.
- One AP with a sector antenna on building A with 1 AP on each of the other 2 buildings with directional panel antennas (maximum gain legally certified antenna) on the other 2 buildings
- Two APs on building A on different frequencies, one pointing at each of the other 2 buildings
- One AP on the building A pointing to one AP on the closest building with a second AP then relaying to the farthest building.
Which option depends on how the network is going to be used, which direction is the traffic flowing to, and how much bandwidth do we need through each building. We also have to understand what the modulation rates are going to be at 1 mile links and 2 mile links. There may also be Fresnel zone issues or obstructions between buildings that may limit my options. Keep in mind FCC legal antenna requirements that may not give you enough gain to overcome a bad Fresnel issue. If for example, you used option 2, you may get MSC(15) rates from A to B but MCS (9) rates from A to C. The MCS(9) rates are clearly unacceptable and may come with a higher packet error issue. That might force you into option 3 to keep rates up. At the same time, you have just increased the amount of traffic through the A to B link. It was this type of situation that forced me to design the network using 3 high-bandwidth licensed options and using twenty 5.8GHz APs to feed into these. I used the maximum amount of the referring party’s equipment without compromising the integrity and price model of the design and still met the required specifications of the system. I also created redundant paths using the 5.8GHz radios to connect the entire critical infrastructure. This supported use of additional 5.8GHz APs which were cheaper than the licensed radios for backup.
The other part of this design utilized a technique that seems to becoming more common. Don’t take manufacturers suggested deployment strategies as gospel. There are many different ways to deploy a radio regardless of the original concept. Nothing says that a mesh radio has to be deployed in mesh designs only. Nothing says that a municipal deployment has to be mesh as demonstrated by TriadLand. There are many ways to deploy the various products out there. For example, I used mesh equipment to design a system that really ended up as a 98% PTMP system. The difference was that instead of 10 APs sharing 100Mbps, I ended up with 10 APs sharing 300Mbps. This is a huge difference.
Now I’m going to step into the shoes of some of you who have called me about projects. Let’s pretend I’m Joe IT manager (yes, I’m being sexist but I don’t have time to make this completely gender neutral and it screws up one of my points later) and the City Manager just came to me and said he wants WiFi downtown. He wants to make the city look progressive and it may enhance some of the business opportunities. More business means more taxes, bigger city budgets, and the City Manager’s name on driving this type of project. However, as the IT manager, my experience is in wire line infrastructure and I’ve never installed anything more than a few APs around the building.
To add to the pressure, as much as a successful project makes careers, unsuccessful ones do just the opposite. I know of IT managers that have left cities after municipal projects failed. Therefore, as the IT manager, I have to first make sure I define what the City Manager wants. Do they want ubiquitous coverage through most of the downtown area so that everybody can connect everywhere? Does the city want to charge for the service or offer it for free? Does the city want to offer connectivity to local businesses that may be limited to one incumbent local internet provider? These are just a few of the questions that have to be asked to the City Manager before the project even starts. They need to clearly define the expectations before you take the project even further. It’s kind of hard to hit a target when you don’t know what it is. If it isn’t in writing, I wouldn’t go one step further as there is no way to be successful if I don’t know what the goal is. There is no defined measuring stick. It’s also a good way to let mission creep into the project and you will never hit the final goal.
In most cases, the city manager usually needs some guidance to help define the goals. The goals may not be within the budget range and compromises might have to be made. If you have helped them define the goals, when you have to go back and tell them the budget restraints you will have more support. In other cases, there might be other departments such as city utilities that might have to be involved. Knowing that up front and having a City Manager mediating between departments is very helpful. Let’s just say that although some departments work great together in the best interest of the city, other departments are well, kind of like asking Randy Johnson to let you get a hit.
Once I’ve defined the goals of the system, then you need to start bringing in people that have that level of experience. Some of that may be in-house, some of it will have to be brought in from the outside. This is where the penny-wise and pound foolish idea comes in. It’s also where many of us computer people don’t want to admit to the people around us that we really don’t know everything. This is where thing may start going awry. Having someone with experience who has been through the war of early deployments can save a lot of hassle and expense later.
We all have a bias towards something. Ford Mustang people would rather eat dirt than drive a Chevy Camaro. Apple users would rather give up their latte’s than use a Windows computer. We all gravitate to our comfort zone. It’ also why many companies use attractive salespeople (all you IT guys know exactly what I’m talking about), have promotions, and take many an IT person to lunch to open the door and help create that product loyalty. Once that loyalty is established, it’s hard to get people to change their mind. Try to tell a Cisco person that there may be a better router product for this application and you might as well have asked them to give up their first born. They will look at you like you came from another planet. Obviously I’m stereotyping in the interest of humor but there is a significant amount of truth in there.
Brochures, free demo units, and great advertising shouldn’t be the criteria used for a deployment. It should be what fits for the project. For example, if you only have 1-2 wired connections to a 1 mile square deployment, you are going to have lots of hops. In that case, depending on the design, you really need 2 backhaul radios in an AP to minimize the bandwidth loss. If there is only 1 hop, it’s not that important. That changes though if a radio is a central point with several APs feeding into it. A 100Mbps AP handling 4 other AP chains under full load is limited to 25Mbps per chain. That’s where a 2, 3, or 4 radio backhaul design becomes very important.
If you want indoor coverage, there is a lot of vegetation, high vehicle traffic load, or guaranteed connectivity to even the lowest powered WiFi devices, then you will need a lot of access points to stay within a short distance of the target client. Having fewer APs which have to go farther also mean more noise is picked up since you are capturing a larger area. That also means that lower powered devices will have trouble connecting. Several poor quality connections will also cause the AP to slow down significantly. There is a saying that a happy customer will tell a friend, an unhappy customer will tell 10 friends. In the world of Internet, an unhappy customer may be telling thousands of friends.
Make no mistake, even if the service is free, it’s public and reflects on your city. If you put up a system that does not work correctly for its target client, that’s worse than if you didn’t put up any system at all. People will try to make the system do things it was never designed for and when it fails, they will make that public. For example, the system gets turned on, the local newspaper writes an article, and one guy can’t connect because he has some old, early production 802.11b laptop sitting directly behind his refrigerator in his apartment. That’s the guy who will be posting in the comment in the newspaper under the Internet article how the system doesn’t work and then will be calling City Hall. Forget the 100 people who are using it successfully, you will never hear from them.
All this information and we still haven’t figured out what AP to use. This process is simply to start weeding through which equipment might be used based on goals and design limitations. We didn’t even cover site-lease agreements for private property, insurance issues, Internet Gateway points, security, zoning, and a few other details which can derail a project. Some of this isn’t directly tied to the AP obviously, but it’s a good idea to be working on this parallel to the design. For example, if I find out that none of the electrical poles have power during the day, then my design changes drastically. Instead of a mesh design, I may be looking at a lot of hub and spoke designs with longer range. Either that or I have to find 20 hour battery backup units for each pole. If I need private property, then I have to worry about aesthetics, 24 hour access, etc… Designs can change depending on many factors. However, these factors dictate which equipment is to be used. Get through all that, re-evaluate the checkbook, and then we can figure out which APs to start looking at.