Tales from the Towers, Chapter 31: When in doubt, use your W.I.T.T.S.

Most WISP operators perform additional projects outside the scope of just providing Internet access. The skill sets required to install backhaul systems to residential or business clients can be utilized in other applications.  Although I’ve been fortunate enough to see many unique types of wireless applications, some stick out more than others. Video surveillance, SCADA for traffic, water, and sewer systems are standard applications, but last week I ran into a new one that really piqued my interest as a wireless engineer and as a guy who regularly reads www.physorg.com.  Although I only understand about 1/1000 of what I read, it’s still really cool stuff.

The trading market is a very dynamic environment.  Trading companies move a lot of shares of stock very quickly based on algorithms that attempt to predict future trends.  Designing these algorithms is apparently the work of math and physics PhDs along with nuclear scientists.  This means that a lot of data centers are stuffed with the fastest, meanest hardware on the planet.  If Intel is thinking about a new processor, these companies already have their orders placed.  When microseconds count and millions of dollars are traded in a blink of an eye, coming in second means more people added to Obama’s unemployment list.  He who trades first, wins.  It’s so time-critical in the data centers that a server that is 60 feet away from a mercantile server would have an advantage over a server that is 140 feet away.

Most people think fiber optics run at the speed of light. Even though Einstein says you can’t make something go faster than 299,792,458 meters per second (the speed of light in a vacuum), he didn’t say it couldn’t go slower. Fiber optics actually run at about .66-.68 times the speed of light (as a comparison, RF runs at about .87 times the speed of light over coaxial cables. It is slight faster across open air.).  The signal doesn’t get a straight shot; it bounces around like a Superball in a basement. It also has to pass through plastic or glass and that slows things down. The result is that if I get to bet on wireless or fiber, for less than a 100 miles in either a flat part of the country or on mountain top property, my money is on wireless.

I love racing.  I’ve raced cars, motorcycles, and my brother for the last meatball in the spaghetti bowl. When given the chance to race the speed of light, bring on the Diet Coke on and warm up the cable modem, I’m going to work.  This is kind of a once in a lifetime problem to solve so there is no way I am going to pass up at least trying to find a solution.  Thus began the foundation for the Wireless Integrated Trading Transport System (W.I.T.T.S.).

The first question is establishing the absolute fastest time that light can travel from point A to point B in a vacuum.  At 30 miles, I calculated that it takes 161 microseconds which means 322 microseconds round trip. Now that we’ve established the absolute floor, my Star Trek mode kicks in. Since Sub-Space is off the table as far as I know, a few other semi-science fiction options start jumping into my imagination.  I remember reading a little while ago about an experiment that pushed radio waves past the speed of light.  The original article I read was on the Los Alamos site but the link is gone.  Here is one that gives the general idea (see article entitled Experts Create Radio Waves Traveling Faster Than The Speed of Light). That idea went down the tubes when I discovered that compressing the wave distorts it to a level that makes it difficult to use for transporting data.  It may yet have a future but I can’t find any other research at this point so it may have gone dark.  I’m also thinking that if we integrate it with some other really amazing technologies, it may be able to do more.

Not to be deterred and wearing my “May the F=ma be with you” t-shirt, I moved on to quantum entanglement.  Ahh, this expands out two options, fastest computers on the planet from quantum computing, and possibly parallel data on the other side simultaneously.  At this stage in the quantum computer industry, I realized that the insurance firm that insures my business was not going to like a cryogenics system in my office. I’m sure I would have to get a rider on my insurance policy attached to a science book. Since it also wasn’t the problem I was being asked to solve, it got tabled to Phase IV.  There are definitely products already available that I need to keep in the toolbox for later but it’s a tangent at this point.  After I’ve squeezed every single nanosecond out of the connection, I’ll come back to this.

Moving back to quantum entanglement, I found some information on Chinese researchers who have successfully transported data 10 miles at a speed of 10MHz with 89% accuracy. Very cool but in reality, they had to transport the entangled photons through the atmosphere using a pair of telescopes.  Keeping the particles entangled is where the problem currently exists so I scratched this one off the list also.  I do see a different application for this that would also help in the project.  However, that technology is not useful today, even with my Snickers Bars budget. I’m putting that into a Phase V category for when I win the lottery.  89 percent accuracy is also about what 80GHz radios are going to do in a snowstorm at 3 miles so it may yet be competitive someday.

I was quickly running out of fantasy technologies.  However, recent headlines concerning neutrinos exceeding the speed of light got my attention. The Opera team at CERN and Gran Sasso think that Einstein might be wrong and published results suggesting that neutrinos exceeded the speed of light. However, the Italians from the ICARUS team recently sent them a congratulatory plate of lasagna with a nice note attached, “Sorry boys, you need to recalibrate your clocks”.  That’s okay though, since trying to add a supercollider on my insurance rider along with the cryogenics system was going to be even harder.  However, we do know that neutrinos go through just about everything, even FCC incompetence and influence, so when the technology eventually does come out, even Julius Genachowski may not be able to regulate it unfairly.

Now that my “blast the competition” options were gone, I came back to earth and started looking at the problem more realistically.  Basically it’s data center to tower, probably half a mile or less, 30 miles to the next tower, and then another half mile or less to the office building.  A simple 3 hop system sounds pretty easy but when every single microsecond counts, even cable and interfaces need to be reviewed.  Although I’ve taken it all the way to the computers in a complete W.I.T.T.S. system, I’m going to stay focused on just one part of the wireless portion here.

Since the most difficult phase of the project is the 30 mile hop, let’s start there.  The natural inclination for most of us is to start with the fastest radios in the industry.  Unfortunately, those are at 60GHz and up.  Some of the bigger players in that area that will go the farthest in the 80GHz range are Bridgewave, GigaBeam, and E-Band.  E-Band and GigaBeam are layer 1 radios so their response time is pretty impressive, in the 4-5 microsecond range per radio.  This is probably per bit but that distinction isn’t in the documentation.

However, we have a conundrum. Speed is the highest priority which is why 80GHz is so attractive but it can’t penetrate a used Kleenex, let alone a snowstorm at any distance. For example, on the East Coast, 80GHz has a maximum range of less than a mile for 99.999% reliability due to weather and atmosphere.  If we have to go 30 miles, the radio salesman is going to be making his quota for the quarter. Of course that’s also if we don’t want to test whether Reed-Solomon really knew their linear algebra. Moving the radios out to 2-3 miles to reduce the number of radios we need to go through, means we either push the error-correcting or we add more delays with retransmits. Retransmits are going to be killers and the farther we go, the worse it gets in all frequencies.

In the radios themselves, Reed-Solomon correction was great when the Beatles were still practicing in Liverpool but it’s dated for this application. In this case, dated means time delays which means losing to my competition. The problem with Reed-Solomon is that while it may be fast enough for the real world, when you are fighting the speed of light, Reed-Solomon is like bringing a Hyundai to run against a Shelby GT500 at the drag races. Although Reed-Solomon is designed to correct errors on the fly, it’s not the most efficient way to achieve maximum performance because of the extra bits required to make it work. Those extra bits have to be transmitted for each packet and every bit means more time, and time is money. I understand why vendors use it since it requires less processing power but here we are playing with the big boys. At the speeds these radios operate, other than gamers, most of the world isn’t going to ever notice a few extra microseconds here and there. Against other trading firms that think the Yankees don’t spend enough on payroll to win, any delay is not going to be acceptable. It looks like this is going to take additional examination which we will get into in a later article.

When you are playing with the limits of physics, the smallest detail becomes important. You need to dig into every nook and cranny, and not accept anything at face value. You also have to throw away conventional wisdom to be the best. The W.I.T.T.S. system came about from that analysis and more. Yes, I know it doesn’t have a lot of value in the WISP industry directly unless 80% of your clients think Call of Duty is reality and going to work is the fantasy, but still it’s fun to see how far our knowledge can take us. Now where’s the TV remote? It’s time for a Deep Space Nine marathon.

* * * * * *

Previous chapter – Tales from the Towers, Chapter 30: Fear of 2.4GHz for public safety defies logic

Next chapter – Tales from the Towers, Chapter 32: S.P.I.R.I.T. – You Can’t Always Get What You Want