As a business owner, I’m behind the basic concept of lower taxes and lower regulations which should generally be the realm of the Republicans. I also like their position supporting the general concept of Usage Based Billing (UBB). On the other hand, they also want to auction off every bit of the available frequencies to corporations. As a small WISP operator that could be bought with Verizon’s office coffee budget for any major city, my chance of winning a bid against them is about the same as Bill Gates buying an Apple laptop.
The Democrats want a neutral internet with no UBB options. That’s good for NetFlix but bad for WISP operators that have limited bandwidth and last-mile resources. It doesn’t matter that Verizon, Sprint, and every other cellular carrier with the same problem are already surcharging for bandwidth usage. However, the Democrats at least leave open the option for future bandwidth being allocated in a similar manner to current unlicensed bands so WISP operators have some growing room. This is kind of like being in the middle of the Amazon River with piranha swimming around deciding how many protein points my toes are, highway bandits on one bank, and a team of transplant surgeons on the other bank waiting to harvest some of my organs.
A long time ago (I’m too lazy to do the research), the government allocated a whole bunch of bandwidth for VHF television, much of it for experimental reasons. The justification was that it was for the public good. During WWII, they did it again with the concept of it was for the public good. Apparently the FCC and the politicians forgot we elected them to do what was right for the public, not what was right for their re-election. Considering that WISPs covered areas that didn’t make a profit for wired services (hence no lobby or campaign contributions) and provided a better and cheaper service than any of the cellular companies for rural areas, that should be enough for them to be recognized.
I stopped believing a long time ago that anyone who gets elected has any interest in helping the public, only in helping themselves keep their job. The budget debate clearly demonstrates that. Since bandwidth across the country that invented internet is basically pathetic compared to many other industrial nations, there should be more interest in helping deliver a better service. Unfortunately, there are still people in Congress who were around when color television was all the rage. They are also the people making decisions on technology that have put the United States years behind other countries to protect their incumbents monopolies (Jim Wright comes to mind here). The best thing for them to do is recognize that they can no longer understand the issues involving these technologies, do what’s in the best interest of our country, and step aside. At this point, even our education system doesn’t care about science or engineering, only protecting union jobs and strengthening political power. I’m not saying that there aren’t good teachers, but even good teachers have to follow the processes that is already in place. Our tax policies have driven manufacturing overseas while half our politicians still think Solitaire is the most amazing Windows game of those that actually use a computer. This does not bode well for our future.
Okay, let’s jump from not understanding to why voters keep putting these same people back in office for decades and get into not understanding RF. Way back in chapter 1 we discussed the idea that WDS connectivity between APs can be a replacement for mesh at a much lower cost. The whole idea of Guerilla Wireless was that a basic $10,000 system or less using a single AP with an omni-antenna could cover a square mile. I didn’t have a site set up for a 2.4GHz test but I figured a 5.8GHz test would give me some idea of the performance of 802.11N and the dual-polarity antenna from Ubiquiti and I had a site ready-made. This wasn’t as complete a test as I wanted but hey, it was valuable information if I wanted a dual AP setup with 5.8GHz backhaul, which is a common configuration. Yes, it’s odd that this configuration with omni-directional antennas fascinates me but I believe it’s the basic building block that re-opens the door to a profitable muni-wireless system. If designed in from the beginning, it’s also perfect for redundancy and in NON-LOS environments, relay functionality to extend tower-centric designs. The area that I’m testing with was designed for both redundancy and low-cost. Due to the remoteness of the area, it was originally conceived with the concept of having 2 locations with separate T-1 feeds at the 2 AP locations being visible from most client locations. When CPE’s are installed, every effort is made to make sure that each client can see either AP. The relay functionality comes in with houses that are at the base of the hills and may only have LOS to the top of the hill.
This area covers about 10 square miles with a hilly environment. They system was originally installed as a 5.8GHz 802.11a system with two base stations. The base stations were 2 houses 1.1 miles apart. For design reasons (okay, it’s what I had in the back room), I installed a 15dBi omni on one house and a 12dBi omni on the other and used WDS to link them together. The AP radios were Ubiquiti Powerstations set to 24dBi output. Connection quality between them was about -77dBi at 12Mbps with 802.11G. This was the highest modulation rate I could get with no packet errors. All users were then connected to one or the other with most users able to jump between them. The three farthest users were between 1.7 and 2.1 miles away. Since Ubiquiti released the first dual-polarity 5.8GHz omni antennas, I figured this was a good idea to test them side by side along with upgrading the area to 80.211N.
I replaced the 15dbi collinear antenna with the 5.8GHz 13dBi dual-polarity antenna and Rocket M5. I then moved the 15dBi antenna, attached a Bullet 5M to it, and replaced the 12dBi antenna. Power output on both radios was set to 24dBm which matched the Powerstations. Immediately there was a 5dbi improvement which theoretically should be about 1dbi and that was with one side being vertical. On the client sides where I had legacy equipment, I saw increases to the 15dBi antenna of up to 13dB and increases of 3-8dBi on the dual-polarity omni. A couple users showed an increase of 1-2dbi but a huge increase in modulation. Across the board, the CCQ% jumped from low 70’s in some cases to 95% or better. That meant that packet errors had dropped to zero on every link.
After I get my Ouiji board out, consult with a palm reader, burn some chicken feathers, and offer Jobu a cigar and rum, I might be able to figure out what happened. There are times when RF is closer to magic than a real science and this is definitely seemed like one of those times. I’m sure there are some logical explanations, so let’s think through them and see if we can figure out what happened.
Between the2 APs, the biggest difference wasn’t only the signal improvement; it was also the modulation levels. Even though one AP had 2x2 MIMO and the other side was 1x1, the connection speed on the vertical polarity was MCS(6) rates or 58.5Mbps. Since I know that to achieve these rates, both the Rocket 5M and the Bullet 5M were now probably running 21dBm output, the increase in signal was even more interesting. Since the signal level is now -72dBm and the previous output was 24dbm, the total increase was 8dB. Throw in the fact the antenna upgrade should have only increased the signal by 1dB, that’s still a net increase of 7dBm. Changing from 802.11a to 802.11n with a 2x2 MIMO signal on one side also had something to do with the number. This area is a bunch of hills that slope at 45 degrees and I’ve seen some interesting phenomenon with this also. Users that have no LOS whatsoever behind a hill and are within roughly a 30 degree deflection angle off of another hill are able to connect. It’s not the primary connection but it does work as a functional backup if necessary. This means that at lower frequencies in some hilly environments, 900MHz might work with 2x2 MIMO for bouncing through the hills. Of course, this also might simply be that the hill has to have this right mineral content, no vegetation, and be in alignment with Venus to work.
The real question is assigning a value to this information. If we go back to Chapter 1, we now know that in a LOS environment where we are using a dual-frequency setup for a wireless deployment, 2.4GHz for area coverage and 5.8GHz for backhaul, it’s important to keep the APs within 1 mile to maintain maximum modulation for each hop. This doesn’t take into account any noise obviously, but this system isn’t designed for Times Square either. Municipalities that want to improve efficiency or security are going to need as much bandwidth as possible, preferably at the best value. We have also improved the efficiency of a profitable wireless system with a minimal Capex for startups.
The dual-polarity omni can also add 1-2 dBi for mobile devices since angle of radiation in a mobile phone isn’t exactly vertical. Polarity isn’t a big deal with a low-gain small antenna, especially when it’s used in more positions than a Yoga class, but every little bit helps. With laptops, it could be worth 3 dBi. However, and this is the cool part, vegetation and irregular attenuation now just became a little bit more porous. Throw in the fact that reflected signal is now multiplied in both number and angle, the signal is going to find a few more holes to maintain the connection. Nothing explains all of the differences and reflections that I’m seeing in one area but dual-polarity and 802.11N are definitely a huge improvement over 802.11a and a single polarity. It also improves the fade margin problem of a single vertical antenna connecting to a low-gain mobile antenna. We now have another tool in the wireless toolbox and it works pretty well.