Thursday, January 24, 2008

700MHz Spectrum: Not So Great For Internet

Today is the beginning of the 700MHz auction, and it got me thinking about the dynamics of the whole thing, and how it reminded me a bit of the rush to blanket the country with wifi. Remember a few years ago when everyone was thinking how good an idea that was? I remember thinking to myself, "am I missing something, because this sounds like a loser to me."

Wifi barely goes 100 feet with any kind of decent signal and so to light up a city with wifi would mean having a wireless router every few hundred feet.

Ridiculous indeed.
Failure guaranteed.

How could so many people sink so many hundreds of millions of dollars into something that obviously couldn't work? I don't know either. But the idea was so compelling, that huge corporations ignored the seemingly obvious.

Well here we go again with 700MHz. Folks are waxing poetic about the significance of this new spectrum to the future of wireless Internet services.

The problem with the 700MHz spectrum is the opposite problem of wifi. First, a lower frequency signal means it carries less data per second. Second, signals in this frequency range go a long, long way, through walls and across miles. This is great for one-way signals. After all, that's how current analog TV is broadcast. But while going only a hundred feet or so isn't very efficient for Internet signals, going many miles with a strong signal isn't great either.

The technical issue: the farther your signal goes, the more data one needs it to carry -- because it must serve more people. Imagine a 30-megabit Ethernet cable being expected to serve 10,000 people. Good idea? Not so much.

Ok, so you say you can solve that by having more towers in a given area so some people can talk to one tower and some to another, splitting up the load. But the problem is that 700MHz signals are so strong that you can't put multiple towers close too each other, because they interfere.

Don't believe me? In the US, our GSM cell phones already operate in both the 850MHz and 1900MHz bands. Some areas have 850MHz service, some have 1900mhz service. 850MHz is low frequency also, and therefore a close comparison to the 700MHz spectrum (particularly since the 700MHz band really operates at closer to an 800MHz frequency).

Riddle me this Batman. Does any cell carrier use 850MHz as anything other than filler (if they use it at all) in urban areas? Nope, because it carries *less* data and you can't have as many towers. Uh Oh.

(NOTE: if you really want to understand the technical issues, read this article on gigaOM. The best part is the comments from people with radio engineering backgrounds.)

So why would all these non-carrier companies (like Google) be getting all giddy about 700mhz for data? Answer: the same reason people believed they could light up major metropolitan areas with ubiquitous wifi.

In other words: I. Don't. Know.

Actually that's not quite true. Google is a special case which I do understand. Google just wants to shake things up and open up the market (which they've already succeeded at). They *ain't* trying to win. And they aren't trying to be a carrier. They *are* smart enough to know that actually buying this spectrum for anything other that voice traffic filler is a loser.

The bottom line is the 700MHz spectrum works well for the incumbent carriers, for which any spectrum is good spectrum. For the other non-incumbent entrants -- not so much. Actually, the fact that Frontline dropped out is some reflection of sanity around the issue. Of course Paul Allen is still in it so there is still quite a bit of err... potentially misdirected optimism. It will be interesting to watch this horse race.

12 comments:

  1. I saw your link on Engadget Mobile's coverage of this and your views on it were quite a nice fresh take and well written. Good work.

    ReplyDelete
  2. That'd be MHz, not mhz...unless you're talking about really longwave stuff.

    ReplyDelete
  3. Thanks much RFDude. I am fixing straight away. With guys like you around who needs and editor ;)

    ReplyDelete
  4. Lower RF frequency doesn't mean fewer bits per second (except in the extreme where the bandwidth approaches the RF frequency). The signal bandwidth and signal-to-noise ratio are the important parameters that constrain bit rate.

    http://en.wikipedia.org/wiki/Shannon%E2%80%93Hartley_theorem

    After looking at the WiMAX (IEEE 802.16) specifications (PHY and MAC), including the use of adaptive antenna systems, maybe you'll be a bit more positive about the 700 MHz possibilities.

    ReplyDelete
  5. Another RF Dude,

    I would agree that WiMax technology can indeed increase the capacity of the signal, and most of my argument is not based on the fact that 700MHz carries less data (though for most in use encoding schemes it does), but the fact that the towers will be so much farther away from each other.

    All of the planning and discussion I have seen around this is based on really substantial distances between towers. The fundamental issue with wireless (including wimax) is that there really isn't enough bandwidth for the things we envision doing with the bandwidth almost no matter what. You can get more efficient, but not more efficient enough. The only answer is cells that are closer together serving a smaller number of people. Adaptive antenna technology can help, but it is obviously not new, and has not allowed 850MHz signals to be used effectively in urban areas.

    I will never say anything is impossible, but given the kinds of expectations that will be placed on any kind of national wireless network I dont think 700MHz + wimax + adaptive antennas get us anywhere near the capacity that will be required in urban areas. I think higher frequencies will be required.

    Interestingly, and totally coincidentally, Cringely covers the same area and reaches the same conclusion as I do.

    http://www.popularmechanics.com/technology/industry/4246037.html

    But I would love to be wrong because I'd love to have lots more wireless bandwidth.

    ReplyDelete
  6. Hank wrote...
    The fundamental issue with wireless (including wimax) is that there really isn't enough bandwidth for the things we envision doing with the bandwidth almost no matter what.
    --
    I agree with you completely. No matter how much bandwidth is available, it will never be enough. New applications will be sought that require more and more information. Imagination and curiosity have no limits.

    Telecommunications convergence will provide a heterogeneous network for "always connected" communications. This will included several RF frequency bands optimized for their bandwidth availability, FCC power constraints, and propagation properties (among other considerations). This will drive the network topology.

    As you rightly observe, no one frequency usage is going to provide all the connectivity everyone wants. Fortunately there are more options.

    Sure there's going to be lots of hype. When isn't there? Hopefully bandwidth usage will be kept open and flexible enough so that new innovations can filter down to the consumers in a timely manner. Our telecommunications and Internet access is already behind a lot of the developed world and we should be leading.

    ReplyDelete
  7. Here are several reasons that 700Mhz is a good choice.

    1) Propagation - In rural America, where ~44% of the population lives, it broadcasts farther, so fewer towers (less cost) and better ground hugging functionality means more complete and better coverage.

    In population dense areas, you can reduce the transmit power to increase the number of cell sites to enable more users per square mile, still at less cost per user. A current cell tower costs rough $500K to $1 million and the better propagation means not missing customers with fewer towers.

    2) Bandwidth - 48Mhz can be cheaply modulated to provide 5 bits per hertz, delivering 240Mbps per tower per sector. The reason I mention sectors is the option of not using omni directional antennas means multiple sectors of operation can be used in congested areas and single sectors in rural areas.

    I have built it before ... I'm sure it would work again. You guys are missing the forest for the trees. He who owns the access to the customer has value ... can anyone say local exchange carrier (LEC). Their value is the connection to the end user, not the services they provide.

    Google owns the most dark fiber in the US. If they purchase the 700Mhz spectrum, they could connect their fiber to their customers and bypass the LEC and that is, in my humble opinion, what they are setting about to do. Open markets ... ha! Google is a business that is trying to reduce the cost of access to their customers.

    ReplyDelete
  8. Hank,

    Your story makes you look like a total wireless noob. It seems like you've done a little research, and then forged on to some mistaken conclusions.

    Your main premise, as you state in your response to "another RF dude", is that the problem with 700MHz systems is that the cell towers NEED to be so much further apart. But they DON'T NEED to be further apart, they simply have the option - which makes covering rural areas much, much cheaper, and given that our country has a lot of suburban and rural populations, this is critical for a national network's profitable economic prospects.

    Here's the simple thing that any wireless network planner could have told you: You can tune down, power down, reduce height, aim any of your 700MHz towers in order to deliberately reduce their range, in order to get higher capacity. (BTW...duh!)

    As a wireless professional, I know that EVERY cellphone carrier does this in urban areas, as they add in new cell sectors, microcells, and picocells. They will often re-aim, re-caliber, and tune down their existign towers. Problem solved. They very often solve it by simply aiming their antennas downwards a little. This keeps the power high, and gets better in-building penetration, while reducing range and co-cell interference.

    BTW, 700 also has much better wall penetration than higher spectrum.

    Oh, yeah, and you sure can get a lot of data through a 700MHz spectrum. Where in the world did you get the idea that throughput is a problem? Sure, 700MHz would get a third of the throughput of 2100MHz IF both systems used exactly the same modulation scheme. But modern modulation schemes (not only WiMAX) can get plenty of bits of data through with each Hz. And a stronger link-budget (provided by better range and wall penetration) means that the network operator can employ more aggresive modulation schemes. I suggest you wikipedia the term QAM (for quadrature amplitude modulation) and OFDM.

    Man, is your article ever wrong! Well, at least you were right about Muni Wi-Fi.

    ReplyDelete
  9. To RFDude & Hank:

    You said "Lower RF frequency doesn't mean fewer bits per second...The only answer is cells that are closer together serving a smaller number of people. Adaptive antenna technology can help, but it is obviously not new.."

    The problem with sub 1 GHz band's better penetration and longer range is not just interference but less ability to differentiate signals, particularly using small, fractional wavelength antennas needed for small CPEs and mobile devices.

    The wireless link of both CDMA/WCDMA and OFDMA technologies are within tenths of a bit per hertz of spectral efficiency. Most now agree that OFDMA has a few 10ths advantage but this is almost irrelevant. OFDM/OFDMA (and variant SC_FDMA used in up-link of LTE) has become the choice for next generation systems because of the simplicity and adaptive capability of using adaptive frequency domain modulation. This is magnified using MIMO, AAS antenna technologies in granular cell topologies.

    And that is what is lost in sub GHz where the signals travel farther and through stuff and thus provice less multi-path signal differentiation. The perceived problem of earlier technologies, this has been harnessed to send multiple signals or enforce signal integrity/reduce interference.

    I have heard that Verizon is trying to press IC, device (and systems) vendors to provide handsets that use MIMO but that vendors have responded that they have not found a practical way to accomplish the feat. Packing multiple fractional wavelength antennas into a reasonable size mobile device does not yield the needed differentiation.

    ReplyDelete
  10. Little late, but this whole article rests on the fact that 700 Mhz is a "bad idea" for dense urban areas. You poor close minded fool...

    Dense urban areas don't need 700 Mhz. 700 Mhz is for farmers, small towns, travelling on the roads, not for places where high-speed data is ALREADY ABUNDANT. They're bridging the final gap. When that get's bridged and everyone has data, then we look at the next step.

    If you haven't noticed, WISPs are popping up everywhere! There's a reason, it's a large market still. I've worked for a couple WISPs recently, if you're not sitting in a flat plain or ontop of a hill, or within 4 miles of a tower, you probably can't get service from them... if they had 700 mhz, ya, everyone would get service, no question.

    Good thing people like the OP aren't making decisions in our society. We'd probably still be using a string and 2 tin cans to communicate.

    ReplyDelete
  11. Just to further my point, think of it this way.

    Everywhere in your nation, you have a consistent 500kbps data connection. Not much right? But the point here, is that it's consistent. That means, everyone can have internet and gps built into their vehicle - this might start to sound scary... but this is the future. Your possession get's stolen? It has built-in gps/internet and as long as the serial is registered to you, and you file a police report - police will be deployed to recover your possession within minutes. Possibility for abuse? Of course, so we create additional layers of security to protect us from ourselves.

    It's a scary future, but I, for one, will embrace it and I plan to make a boat-load of money from it.

    ReplyDelete
  12. The lack of available frequency is result of the what FCC has managed, disected and sold the frequency by blocks. Had they approached it as contiguous bands partitioned by power intead, as done though once unclassified TimeDomain PulsOn technology the story would be much different.. But then again whos to say the gov isn't using it?......

    ReplyDelete