Saturday, August 13, 2016

Wireless v Fiber

There have been some recent moves in expanding wireless. Fiber is still fiber. Let us examine the differences and try to explain to some people what the facts are.

Recently in Backchannel[1] one of the writers, a lawyer I believe by calling, has made statements which in my opinion and my experience are not just wrong, they a truly outright apparent fabrications based on nothing that is in my opinion acceptable to those with even a modicum of competence.

Let me first restate some bona fides. Besides a PhD in EECS from MIT in communications, I then added some fifty years of design and deployment experience in wireless and fiber. One need look no further than a list of hundreds of papers on the topic. I have built out fiber in about twenty countries and frankly found the US the most difficult due to Franchise rules and pole attachment regulations. The incumbents in the US have a permanent barrier to entry for any new entrant. Put that aside for the moment.

Let us first compare fiber and wireless.

I. Fiber

Fiber has substantial capacity. Yet it requires many hurdles and costs an excessive amount per subscriber in capital. Let me list the hurdles:

1. Franchises: In every town and state there are Franchise requirements. You just can't build out a fiber system. You must get permission. The problem; twofold. First towns have Selectmen or the like who generally are clueless, often supported by Cable Companies, and willing to spends months if not years negotiating a Franchise. This adds thousands of dollars to the cost per subscriber and is all too often not realized.

2. Pole attachments: If you get a Franchise, then you have to get pole rights or other rights of way. You cannot start the negotiations until you have a franchise so it is sequential and the incumbents who owns the poles is in no hurry to get to the end.

3. Build Out: The laying of fiber has negative economies of scale. Even assuming the fiber is free, which it is not, the labor costs are always increasing and the delays are ever expanding. What may have cost $50,000 per mile five years ago is now $75,000.

4. Drops: Assuming you have achieved the above then you must get to the subscriber. That is the drop. It generally must be buried and if say you are in New England the rocks etc. will drive the costs to extreme levels.

5. Capital: The Capital per Sub can readily exceed $5,000 which is quite excessive. If the above were non-existent then one could do it for almost a tenth but the above are real. Our lawyer friend seems to be ignorant of these facts.

II. Wireless

Wireless is a totally different tale. First the key difference is the lack of infrastructure. You do not need a Franchise; you need a license but if you already have it so be it. Here are the advantages of wireless:

1. Ever Scalaeble Technology: The introduction of 4G with OFDM allowed the bits per second per Hz to go from 1 to 10. For 5G we see that using multiple beam antennas we can go from 10 times to 100 times! That means each user can get well in excess of 10 Gbps.

2. Capital is Incremental: Unlike fiber and even more so unlike a satellite system, wireless capital per subscriber and be deployed incrementally. I demonstrated that twenty-five years ago! Again we did it. Fiber requires a build of infrastructure. Wireless builds as we follow the customers.

3. Technology Changes in Short Time Periods: Cable TV converters are an average of 10-15 years old. They seem never to be replaced or upgraded. A wireless device is upgraded every 18 to 24 months! Thus the customer can follow the technology curve. As one upgrades cell sites using software defined modems and the like, then technology is always at the leading edge and the capital to the infrastructure provider is low.

4. Distributed Systems Can Evolve: As we build out systems we can do so in a distributed manner. WiFi can be integrated with backbone wireless and mesh networks are readily available.

We have argued again and again that the Google fiber builds were fruitless. Now we see they want to build out wireless. Is Google seeing the light? Not really, they needed licenses. If there is however a sharable band then perhaps they can execute this strategy.

Now for what in my opinion are the falsehoods of this lawyer:

Statement 1 is:

One way to increase the information-carrying capacity of a wireless network is to encode data on those wobbling frequencies more efficiently. The standards you’ve heard about — CDMA, 3G, LTE — they’re all about jamming more data into each unit (hertz) of spectrum. A new 5G set of standards will do the same thing, in an even fancier way: the antennas for very, very high frequencies can be so tiny that you can put 8 or 16 of them into a handset or base station and then have them all work together in an array to create a beam of data. Tons and tons of data can be carried on those aggregated beams. Transmission beams in an array can be steered in milliseconds to point to an individual user. You couldn’t do this kind of thing at lower frequencies, because many antennas would need, say, three feet of space — and you can’t fit that into a handset.

Yes, you can use small antennas at the lower frequencies. Ever hear of Ham radio? I have a 140 MHz hand held set, I can create a beam from a set of small antenna. Ever hear of WiFi, even 802.11n uses MIMO, many antennae. The above statement is just wrong. But even more so, the real antennas are beam-formers at the base station! I did this in 1992, and filed it with the FCC for my Pioneer Preference. We developed it jointly with MIT Lincoln Lab. The military has done this for decades. The statement as presented is just wrong, totally wrong!

She continues:

Until there’s a standard, carriers that want to be able to reach global markets won’t be anxious to make devices that will work in just a few places. They want to be able to use the same frequencies everywhere. Current phones and other widely-used private-sector communications devices have radios that transmit and receive only frequencies below 6 GHz, and the very, very high frequency spectrum that the FCC recently said it would open up for 5G purposes is all above 24 GHz. So we have a huge legacy replacement problem that will take a while to overcome and requires a standard to fix. All of this takes years.

Now back in 1990-92 I was COO of what is now Verizon Wireless. I worked with Qualcomm to introduce CDMA. We worried about turning the ship, from analog to digital, but it worked, seamlessly. Frankly I would suggest that the customer never noticed. Why? Simply because the replacement time for handsets is about 2 years! Thus with such a short replacement time the turnaround is painless. It did not and does not "take years". I did it! Good Lord, look at the facts!

She continues:

Again, wireless and fiber are complementary. Carriers know this. People call the cables between cell towers and central network offices “backhaul,” and when Verizon launched its 4G LTE network in the US covering 93% of the population it needed about 30,000 towers, each one of which had to have a fiber connection. But for a high-frequency 5G spectrum to cover that same population, you’d need to reach many millions of towers and base stations with fiber. Remember, you need to be very close to base stations to pick up and transmit these ginormous amounts of data across high-frequency airwaves. We’re going to have to have fiber interconnection points right next to houses and office buildings, and in many places fiber running inside those buildings. And to reach indoor areas with reliable high capacity, you’ll need multiple antennas inside rooms that can beam signals towards you from multiple angles (to avoid the “people as bags of water” problem).

The back-haul has always been with us. We actually used wireless for many of them but alas fiber can work as well and it can be shared with multiple carriers as are the cell towers.  I agree that with the ultra-high bands proposed for the new releases they are of short distance. Worse they do not work in humid environments, tried that one folks. So using the higher bands may not be really good to use and why then does one assume millions of towers! That is just in my opinion a stupid idea! One may then have an evolving multi-tiered network, with micro and nano cell nodes at customer premises as we have WiFi today.

Also 5G is a technological change, and evolution. One can use the old bands but with the new technology. One can get 100 bps/Hz and with a few hundred MHz and adaptive beam formed antennas one gets Gbps links to local users on demand.

It is my opinion that this lawyer has put forth a straw man that does not in any manner reflect reality. Indeed, no one is proposing building millions of towers. Engineers just are not that, shall we say, stupid. I cannot perhaps say the same for those technologically impaired.