16 Comments

Summary:

The government is hoping that making a band of unlicensed spectrum available as part of the upcoming incentive auctions will help build a nationwide wireless network. Is that the best use of that spectrum?

mastwheat

Recently there has been a push to make a significant amount of unlicensed white-space spectrum available in the 600 MHz band as part of the Broadcast Television Spectrum Incentive Auction Rulemaking. As reported in BNA, the FCC is considering making an additional 30 MHz of spectrum available for unlicensed use, augmenting existing white-space spectrum. Proponents of this unlicensed band are using the term “Super Wi-Fi” to describe the technology that would use this spectrum. The only problem is that it’s not super for multiple reasons, and it’s definitely not “Wi-Fi.”

The term Wi-Fi refers to interoperability compliance with specific IEEE 802.11 standards, and is a designation controlled by the Wi-Fi Alliance, the organization that certifies Wi-Fi gear. The Wi-Fi Alliance is not happy about the term “Super Wi-Fi” had this to say in a press release last year, “The technology touted as “Super Wi-Fi” does not interoperate with the billions of Wi-Fi devices in use today.” In addition, they state, “Wi-Fi is a registered trademark of the Wi-Fi Alliance and the term ‘Super Wi-Fi’ is not an authorized extension of the brand.” So let’s just call it “white-space” network, which is the origin of this technology.

Since it’s not Wi-Fi, it needs new radios

wifi hotspotBecause super Wi-Fi isn’t a new band and has a new radio standard, existing Wi-Fi radios in phones, tablets, and laptops won’t work with these white-space networks. Any user wishing to connect over a future white-space network will need entirely new equipment, most likely a USB form-factor modem. Another possibility would be a wireless router that connects to the white-space network and provides Wi-Fi connections as a hot spot; however, that newly created Wi-Fi hotspot is then subject to all the congestion frailties we currently experience on Wi-Fi networks today.

Another important issue is the radio standard itself. There are two different standards being developed for white-space spectrum, IEEE 802.11af and IEEE 802.22. IEEE 802.22 was just recently completed but IEEE 802.11af is still in development. It’s not at all clear which of these standards will prevail in the market, or whether something entirely new will come along. Dueling standards generally serve to confuse and delay markets.

Now let’s try to understand the “super” part of this technology, since I don’t really see anything that “super” about it. First, it’s quite slow compared to existing Wi-Fi technologies, limited to a peak rate of 29 Mbps. In contrast the latest Wi-Fi standard, IEEE 802.11ac which is still under development but available in commercial product, can deliver throughput rates close to 1 Gbps (800 Mbps) in a base configuration, and over 6 Gbps in its most advanced configuration.

Who will build the networks?

Verizon LTE footprint March 2013Second, claims about its superiority are based on an assumption that as-yet-unidentified service providers will deploy networks to operate on this white space spectrum all over the country and offer wireless broadband service. Policymakers seem to hope that these new networks will somehow alleviate the mobile broadband capacity crunch that we are experiencing.

This notion, however, is flawed. First, it is extremely unlikely that any entity will invest billions of dollars in massive amounts of network infrastructure to use unlicensed spectrum to support commercial wireless broadband services. The carrier’s inability to guarantee service quality, predict and manage capacity, and eliminate or prevent interference render unlicensed spectrum an inferior solution for providers who compete based on quality of service and ability to support bandwidth-hungry apps and devices.

Add to this the possibility of different technologies using this band and it looks like an even less attractive basis for a significant capital expenditure which needs a return on the investment. For example, an IEEE article states that there is likely to be heavy degradation of 802.22 performance if it operates alongside an 802.11af network.

One thing we have learned over the least twenty years of building wireless data networks is that large volumes of users, whether its consumers, business users, or even M2M applications, subscribe to a wireless network technology only if they can obtain really broad coverage. Wireless network technologies with limited coverage have achieved only limited commercial success, including technologies such as Cellular Digital Packet Data (CDPD), Metricom Ricochet, and most recently Sprint/Clearwire’s WiMAX network.

The Google white spaces database in action.

The Google white spaces database in action.

White-space networks will be similarly limited in its coverage, but will further be complicated by being suited only for fixed operations. This is because the technologies currently envisioned to operate in the white space spectrum rely on the modem’s current location to query a database to learn what frequencies it is authorized to use.

Is LTE a better bet for this spectrum?

In contrast, wireless data technologies that have enjoyed wild success, such as EV-DO, HSPA, and LTE are technologies with extremely broad coverage, coverage achieved from tens of thousands of base stations covering almost all of the population. It may seem to be an apples to oranges comparison to compare a commercial LTE network with a white space network, yet it is exactly this comparison that needs to be made, because the spectrum being contemplated will end up being used for LTE networks or for white-space networks. There is no middle ground currently under discussion or development.

I believe one effective basis for such a comparison is to consider the aggregate capacity the two different networks might provide. The math for this is straightforward. Simply consider the number of possible sites, multiplied by the amount of spectrum, and multiply that by the spectral efficiency.

Cell Tower and OspreyAccording to CTIA, there are some 285,000 cell sites. Assuming the spectrum was auctioned, cellular operators would likely deploy the spectrum across most of these sites, but a conservative estimate would be half these – 142,000 sites, each with 3 sectors. Taking 30 MHz of spectrum under consideration and average LTE spectral efficiency of 1.4 bps as per my studies and writing on this topic, that amounts to 17,640 gigabits/second (Gbps) of additional national mobile data capacity.

White-space networks could have comparable spectral efficiency and could also be deployed in 3 sector configurations. The only variable in question to determine the total capacity delivered by white-space networks using the same amount of spectrum is the number of sites (access points). Given my previous arguments of interference concerns, it’s inconceivable that anybody would build out white-space networks with density equivalent to cellular network.

White spaces are for local networks, not national ones

White spaces might be good for coffee shops?

White spaces might be good for coffee shops?

Usages are more likely to be adhoc and localized – just as with Wi-Fi. It might make sense to deploy a white space network on a campus, at an oil well, or in a town, but given the lack of control over the spectrum, it won’t make sense for any entity to deploy a national network. As a consequence, my expectation is that the total number of white space sites will be significantly lower than that of today’s cellular networks, and thus the aggregate national data capacity provided by the use of that spectrum will also be significantly lower. This lower data capacity represents a lost opportunity for the spectrum.

It’s all well and good to create experimental networks and to foster innovation, but the 600 MHz band represents a precious resource at a time when providing sufficient capacity to foster the mobile broadband revolution is crucial, and a time when new sources of spectrum seem ever more challenging.

I believe applying that spectrum to technologies that will use it the most fully will provide the greatest societal and economic benefit. Right now, those technologies include LTE and LTE-Advanced. We should continue to foster innovation and experimentation with white space spectrum and Wi-Fi, but not at the expense of also expanding the base and capabilities of our best-in-class, commercial wireless broadband networks that depend on licensed, exclusive use spectrum for their core operations.

Peter Rysavy is President of Rysavy Research, a wireless network engineering firm.

  1. Thank you, Peter, for the best analysis I have seen on this topic, which seems to have had way “more heat than light” on it recently. After talking with an old Intel WiMax friend about this, the devil is clearly in the details, and LTE might still be a better option The technical issues are just challenges. The market issues are the real drivers. I see this generic capability perhaps having greater applicability in the developing world, where terrestrial network infrastructure is limited to non-existent.

    Share
  2. Reblogged this on mayo615 and commented:
    This is the best technical and market analysis on Super WiFi I have seen since I posted “Free WiFi for the Masses: Devil in the Details” on this blog. It proves that there is a lot of loose talk out there by people who do not know what they are talking about.

    Share
  3. It’s important to remember that the original goal of the white space initiative was to provide free unlicensed spectrum that could be used to deliver cost-effective wireless broadband access in rural areas, where laying cable or fiber is not feasible. Of course, there aren’t that many potential customers in rural areas, so the white space concept was soon co-opted as a means of serving urban customers.

    As you point out, white space is really not all that attractive to mainstream carriers, who must provide a consistent user experience nationwide. The option of bidding for a license to a private slice of spectrum in the 600 MHz band is a much more attractive option for most carriers. The auction may also delay the development of white space spectrum, because no one will know how much white space will be left until after the post-auction repacking of the TV band. Would you invest millions in developing a product or service based on spectrum that might not be sufficiently available in 3-5 years?

    Share
  4. Just want to point out a few facts:

    Firstly, nobody has used WIFI to build nationwide network, it is a big success in commercial, no doubt.

    Secondly, carrier can never deliver ubiquitous wireless broadband, the rural is not in their interests.

    Thirdly, 2.4GHz is free, no one has even thought about to develop a LTE system in that spectrum, as not sure if it will work. If the case, no one is sure LTE will work in this 600MHz spectrum, if it is unlicensed as 2.4GHz.

    Share
  5. I have to say, that while you’re correct in criticizing the label of WIFI. However I think you’re way off the mark on the rest.

    First, several organizations are building networks today. Second, the propagation characteristics of the spectrum do make it “super” from how it can be used. Third, I don’t know who (if anyone) has championed the idea that nationwide networks built on this technology will spring up anytime soon, but that looks like a straw man since the vast majority of the networks are being built either to service rural broadband or support M2M networks.

    Finally, the assertion that LTE would somehow be a superior technology to use the frequency flies in the face of the fact that this is inherently a shared resource and regardless of the protocol winner(s) we don’t have to be locked into a single license holder in an area. Would you call the ISM bands ( 2.4 and 5.8) a better fit for LTE than WIFI? Why is this spectrum different to you?

    Share
  6. Steve Crowley Tuesday, March 19, 2013

    I don’t see how it’s super. Lower frequencies will propagate better through walls, etc. The problem is interference propagates even better than coverage at lower frequencies (each wall attenuates it less), and I don’t think one can match the AP densities that can be achieved with 2.4 and 5 GHz Wi-Fi systems. For some applications when coverage is more important than capacity, fine. Super? In the sense that Superman is better than man? No.

    Share
    1. That’s because you’re only thinking of the WIFI paradigm, which is to be understood since some of the earlier proponents of the technology (mistakenly IMHO) used that to promote TVWS. Think about it like this in a WIFI or really any other data over wireless RF deployment you have to deal with both contention and coverage. Both can be problems and sometimes they are both issues in the same network.

      When Microsoft was initially testing TVWS on their main campus they could cover the entire area from central TVWS AP. Now, of course that single AP couldn’t come close to dealing with the contention from all the devices (assuming they could all magically speak the new frequency/protocol) but compare that to the several thousand WIFI APs that MS uses to cover their campus. Could you provide similar service to the MS campus with a hundred TVWS AP’s? Probably not, but I’d say that you could at least cut the number of APs needed by half.

      Performance is of course the next question. A single TVWS channel is not going to be able to carry as much a single 802.11n or ac channel, but we’re not limited to a single channel. In most places there are many available channels. Look at what, for mobile devices, is available on the MS campus 11 6 MHz channels which can be bonded even though there isn’t gear available yet. Its worth noting that these 6 MHz channels are the same as the ones used by cable operators to provide DOCSIS 3.0 services though I doubt we will similar modulation speeds (you can’t run QAM on TVWS).

      21 512-518 40 **
      23 524-530 40 **
      26 542-548 40 **
      28 554-560 40 **
      30 566-572 40 **
      32 578-584 40 **
      34 590-596 40 **
      41 632-638 40 **
      43 644-650 40 **
      45 656-662 40 **
      49 680-686 40 **

      Share
  7. “super wi-fi” was a silly name and most people have got over it by now.
    LTE is not a panacea! It is optimized for highly mobile applications in exclusive use spectrum. Wi-Fi is almost the exact opposite, mostly stationary and opportunistic access.
    Both have been a great success, for these entirely different reasons. Given that there are more unlicensed devices operating in the world than licensed devices making part of the reclaimed spectrum available for licensed use and part for unlicensed use is more rational than most of the argument made here.

    Share
  8. Localized limited bandwidth network applications like encrypted digital wireless microphone systems would seem to be the best use of this kind of white space spectrum.

    Share
  9. Christopher Mitchell Tuesday, March 19, 2013

    The better question is not one of technology, it is one of economics. Do we want more spectrum locked away for only one or a small number of firms to use or do we want to open it to everyone? Wi-Fi would never have revolutionized wireless if it were limited to a few firms to develop, it thrives because it is a commons that is open to innovation.

    The people trying to prevent new unlicensed spectrum are (often but not always intentionally) defending a duopoloy that is hurting our economy with its high prices and low monthly caps.

    Share
  10. Richard Bennett Tuesday, March 19, 2013

    The basic conundrum for unlicensed is population density.

    There are several initiatives underway to make more spectrum available above 3 GHz, where it makes sense. As Peter points out, unlicensed spectrum below 1 GHz only makes sense in very sparsely populated rural areas because a channel that covers too many users can’t provide any of them with good performance.

    Wi-Fi is a success because the spectrum used in one home can be used by another home 600 feet away. Sub-1 GHz systems with higher power than Wi-Fi can only re-use spectrum every 6000 feet or so. That’s great for the farm, not even helpful in the suburbs, and a complete disaster in the cities.

    In addition, unlicensed media access control (MAC) protocols don’t share access to a common frequency nearly as well as licensed spectrum MACs do. Unlicensed is all based on contention, and contention protocols don’t scale.

    The argument for white spaces seems to come down to “I hate the phone company.” That’s all fine and good, but you need a lot more than emotion to design and build a functional network, and every time the FCC allocates spectrum to a system that won’t work they take it away from those that do work.

    Why waste a vital national resource on poorly thought out, emotional, reactive schemes? We can do a lot better, and we have done better in the past.

    Share

Comments have been disabled for this post