Wednesday, October 16, 2013

Unlicensed’s success: physics, not regulation?

Unlicensed allocations have generated a massive, and to many surprising, amount of innovation and value (see the References below). The question is: Why?

Almost all of the value so far has come in the 2.4 GHz ISM band, mostly due to Wi-Fi but also to a lesser extent Bluetooth applications. There is never a single, simple answer to a Why question about a complicated nexus of technology, politics and user behavior, but my impression is that unlicensed partisans believe that it's due pretty much exclusively to the techno-economic characteristics enabled by the rights assignment regime: “openness” (Benkler), “managed commons” (Milgrom, Levin & Eilat), or “rule-based access” (Thanki).

I think it's at least plausible that Wi-Fi's undoubted success has been due to a fortuitous coincidence of band choice, physics and timing as much as to regulation: It turned out that the interference range was small enough that users didn’t really degrade each other’s performance; and the networking needs of their applications could be met by the bandwidth available around them. In other words: the capacity of the channel was larger than the number of people who interfered with each other, multiplied by the data they wanted to move.


Some key factors were:

  • The combination of 2.4 GHz operating frequency and indoor use limited interference between Wi-Fi access points;
  • The dominant application was multi-media web browsing, which provides excellent user value but is tolerant of some loss and delay;
  • The 3 x 22 MHz bandwidth enabled by the 84 MHz allocation was sufficient to support these kinds of broadband for users within the interference range, in most cases.

Other things also helped:

  • The cost of Wi-Fi infrastructure was largely invisible: consumers bought boxes for home networking, and increasingly got them “free” as part of cable and telco broadband routers; Wi-Fi was deployed as must-have amenities (like restrooms) in public venues;
  • The high volume and relative simplicity of Wi-Fi components meant they were cheap, and became ubiquitous, feeding a virtuous cycle;
  • The explosion in demand for mobile data enabled by the iPhone phenomenon – emerging from the licensed, not unlicensed, ecosystem – could leverage the installed base of Wi-Fi access points in homes and public venues;
  • IEEE 802.11 and the Wi-Fi Alliance emerged as de facto band managers, as argued in David Reed and Jim Lansford's 2013 TPRC paper.

To the extent that unlicensed’s success is due to the particularities of indoor use at 2.4 GHz of multi-media web pages, rather than “commons” allocation, the success in this band won’t necessarily follow automatically elsewhere, e.g. in the TV bands where interference ranges are much larger; for use in outdoor wide area coverage, as contemplated by the cable industry (remember ill-fated Muni Wi-Fi?); or as much more demanding applications like streaming HD video become dominant. However, it is likely that unlicensed will grow from strength to strength in bands and venues where the interference range is small and allocated bands are wide, e.g. 5 GHz and above.

I suspect the wild success of cellular-like licenses is also due to some fortunate coincidences (including the success of unlicensed), but that’s a topic for another post.

References

Some recent paeans to unlicensed:
Benkler, Y. (2012). Open wireless vs. licensed spectrum: Evidence from market adoption. Harvard Journal of Law and Technology, 26(1). http://cyber.law.harvard.edu/publications/2012/unlicensed_wireless_v_licensed_spectrum
Milgrom, P. R., Levin, J. D., and Eilat, A. (2011). The case for unlicensed spectrum. Google. http://ssrn.com/abstract=1948257
Thanki, R. (2013). The case for permissive rule-based dynamic spectrum access. Microsoft. http://research.microsoft.com/en-us/projects/spectrum/case-for-permissive-rule-based-dynamic-spectrum-access_thanki.pdf
This thinking has been was heavily influenced by Doug Sicker et al.'s 2006 TPRC paper that used a combination of US population density analysis and system performance simulation to argue that given a 56 m interference radius and 3 Wi-Fi channels, 90% of population was within interference range of fewer than 20 other people – and that web browsing or VOIP only break down for more than 20 users
Sicker, D., Doerr, C., Grunwald, D., Anderson, E., Munsinger, B., and Sheth, A. (2012). Examining the wireless commons. In TPRC 2006http://ssrn.com/abstract=2103824 


No comments: