Yes, much more work needs to be done but I’m amazed at how fast it IS getting done, and built into consumer electronics and pushed out the door affordably and put into users’ hands…where whole new layers of license exempt creativity come into play.

Tim, in your opening comment you cite Eli Noam’s
seminal article on open spectrum (1995), which moved insights just taking root among engineers onto a political-philosophical plane, where they bloomed faster than the engineering roots could feed at the time. Though Noam is an academic, he is a ham radio operator who knows the technology better than most of his followers, and has professional experience as a regulator, too, (at the NY state utility commission) which hardly any of us can claim. That he has backed off from his early utopian claims for Open Spectrum disappoints many like Benkler and Lessig. Now he argues for “real-time markets” in spectrum access and transferable interference rights – concepts too obtuse for a mass movement to put on placards.

Clearly what’s needed is a more sophisticated approach to Open Spectrum, not abandonment of the hopeful trends it embodies. The fact is that we have accepted draconian restrictions on radio use for a century, because the technology was too dumb and too powerful to leave uncontrolled. We would never accept the kind of licensing restrictions on video cameras and copy machines that seem utterly normal and necessary for radio. But radios ARE getting smarter, more autonomously skilled at avoiding interference and cross talk (when was the last time you turned on your mobile phone and broke into a conversation already in progress – compare that with CB).

And more importantly, the applications that are proving popular are often very low power – the signals don’t even have to leave your home. The combination of lower power and smarter equipment means radio can already be treated with a much lighter touch.

I would argue that there is no justification for a national government to license the use of equipment whose signals are completely confined within the walls of one family dwelling. And it is in fact dangerous to accept such intrusive government power as right and proper. The home is a space where we should be able to implement Open Spectrum ideals right now. Then maybe our children can expand that domain as the equipment gets more clever about managing emissions.

Yes, much more work needs to be done but I’m amazed at how fast it IS getting done, and built into consumer electronics and pushed out the door affordably and put into users’ hands…where whole new layers of license exempt creativity come into play.

Tim, in your opening comment you cite Eli Noam’s seminal article on open spectrum (1995), which moved insights just taking root among engineers onto a political-philosophical plane, where they bloomed faster than the engineering roots could feed at the time. Though Noam is an academic, he is a ham radio operator who knows the technology better than most of his followers, and has professional experience as a regulator, too, (at the NY state utility commission) which hardly any of us can claim. That he has backed off from his early utopian claims for Open Spectrum disappoints many like Benkler and Lessig. Now he argues for “real-time markets” in spectrum access and transferable interference rights – concepts too obtuse for a mass movement to put on placards.

Clearly what’s needed is a more sophisticated approach to Open Spectrum, not abandonment of the hopeful trends it embodies. The fact is that we have accepted draconian restrictions on radio use for a century, because the technology was too dumb and too powerful to leave uncontrolled. We would never accept the kind of licensing restrictions on video cameras and copy machines that seem utterly normal and necessary for radio. But radios ARE getting smarter, more autonomously skilled at avoiding interference and cross talk (when was the last time you turned on your mobile phone and broke into a conversation already in progress – compare that with CB).

And more importantly, the applications that are proving popular are often very low power – the signals don’t even have to leave your home. The combination of lower power and smarter equipment means radio can already be treated with a much lighter touch.

I would argue that there is no justification for a national government to license the use of equipment whose signals are completely confined within the walls of one family dwelling. And it is in fact dangerous to accept such intrusive government power as right and proper. The home is a space where we should be able to implement Open Spectrum ideals right now. Then maybe our children can expand that domain as the equipment gets more clever about managing emissions.

There are many reasons for the range limitation of WiFi:

1) The operating frequency: As the operating frequency gets above approximately 2.5 GHz, RF propagation conditions get pretty poor. Below 2.5 GHz is generally considered the sweet spot for coverage over any significant distance or outdoor-to-indoor coverage (the signal strength tends to drop by a factor of 20, or 13 dB, as it goes through exterior walls). So the 2.5 GHz ISM band is much better than the 5 GHz ISM band. The higher frequencies are fine for outdoor-outdoor links with directional antennas. That’s why some mesh systems are using 5 GHz for the mesh links between the outdoor aces points and 2.5 GHz for the links into people’s homes. So there’s more spectrum up at 5 GHz, but it has poorer propagation characteristics. And the 2.5 GHz ISM band is totally overloaded (my WiFI at home gets noticeably crappier when I turn on my microwave, which is off course a 600 watt co-channel transmitter, albeit a shielded one. In addition, there are cordless phones, garage door openers, etc.).

2) Regulatory constraints: the FCC limits the transmit power of any unlicensed device. For point-to-multipoint scenarios (i.e. an outdoor mesh network access point), the limit is 4 watts EIRP (i.e. the power at the antenna output). In contrast, a 3G or WiMAX basestation’s downlink EIRP would typically be anywhere from 200 watts to 1000 watts. That’s a pretty huge difference when it comes to coverage and building penetration. The uplink transmit power levels of WiFI and WiMAX or 3G devices are closer to each other, which makes sense because they’re all battery powered devices. The transmit power is typically dictated by battery life concerns rather than regulatory limits (though they do exist).

3) Interference: since anyone can transmit in the unlicensed band as long as they meet maximum transmit power limits, there will inevitably be interference as multiple access points operate in the same geographic area. The presence of this noise limits the coverage range of an access point.

A more general problem that’s not directly related to range is multiple access. 802.11 is an un-coordinated system where any device can transmit at any time and as such, it achieves fairly poor ( below 50%) utilization of the underlying physical capacity of its radio spectrum (for the uber-geeky, it’s a variant of CSMA/CD that’s designed for wireless). In contrast, a centralized architecture where a scheduler at the basestation controls user access (e.g. any cellular technology like WiMAX, 3G, 1xEvDO, etc.) tends to get closer to 80 or 90% efficiency. This is one reason why WiFI access points don’t scale well to support large numbers of users.

There are many reasons for the range limitation of WiFi:

1) The operating frequency: As the operating frequency gets above approximately 2.5 GHz, RF propagation conditions get pretty poor. Below 2.5 GHz is generally considered the sweet spot for coverage over any significant distance or outdoor-to-indoor coverage (the signal strength tends to drop by a factor of 20, or 13 dB, as it goes through exterior walls). So the 2.5 GHz ISM band is much better than the 5 GHz ISM band. The higher frequencies are fine for outdoor-outdoor links with directional antennas. That’s why some mesh systems are using 5 GHz for the mesh links between the outdoor aces points and 2.5 GHz for the links into people’s homes. So there’s more spectrum up at 5 GHz, but it has poorer propagation characteristics. And the 2.5 GHz ISM band is totally overloaded (my WiFI at home gets noticeably crappier when I turn on my microwave, which is off course a 600 watt co-channel transmitter, albeit a shielded one. In addition, there are cordless phones, garage door openers, etc.).

2) Regulatory constraints: the FCC limits the transmit power of any unlicensed device. For point-to-multipoint scenarios (i.e. an outdoor mesh network access point), the limit is 4 watts EIRP (i.e. the power at the antenna output). In contrast, a 3G or WiMAX basestation’s downlink EIRP would typically be anywhere from 200 watts to 1000 watts. That’s a pretty huge difference when it comes to coverage and building penetration. The uplink transmit power levels of WiFI and WiMAX or 3G devices are closer to each other, which makes sense because they’re all battery powered devices. The transmit power is typically dictated by battery life concerns rather than regulatory limits (though they do exist).

3) Interference: since anyone can transmit in the unlicensed band as long as they meet maximum transmit power limits, there will inevitably be interference as multiple access points operate in the same geographic area. The presence of this noise limits the coverage range of an access point.

A more general problem that’s not directly related to range is multiple access. 802.11 is an un-coordinated system where any device can transmit at any time and as such, it achieves fairly poor ( below 50%) utilization of the underlying physical capacity of its radio spectrum (for the uber-geeky, it’s a variant of CSMA/CD that’s designed for wireless). In contrast, a centralized architecture where a scheduler at the basestation controls user access (e.g. any cellular technology like WiMAX, 3G, 1xEvDO, etc.) tends to get closer to 80 or 90% efficiency. This is one reason why WiFI access points don’t scale well to support large numbers of users.

“it had been rendered obsolete by technology. In particular, it had been rendered obsolete by the fact that the declining cost of computation and the increasing sophistication of communications protocols among end-user devices in a network made possible new, sharing-based solutions to the problem of how to allow users to communicate without wires”

The reality is that there are two key problems/challenges that will push out a spectrum commons approach for years (or decades):

1) The required technologies are far from being ready. Software defined radio (SDR) and cognitive radio (CR) are the two key underlying enabling technologies and both of these are far from being consumer technologies. SDR is used to build radios that support multiple interface technologies (e.g. CDMA, GSM and WiFi) with a single modem by reconfiguring it in software. However, SDR modems are expensive since they typically entail programmable devices like FPGAs, as opposed to the mass produced, single purpose ASICs that are used in most consumer devices today (and are key enablers for low cost handsets). Even today’s multi-mode devices tend to just have multiple ASICs (or multiple cores on a single ASIC). SDR is currently mostly used in military applications where cost is no object. SDR is also a modem technology and it ignores RF design issues. In particular, the RF design of a wireless device is typically closely coupled with the underlying access technology and modem design. E.g. different air interface technologies have different spectral mask requirements and different degrees of vulnerability to co-channel interference and strong adjacent channel power. A device that must work over a wide bandwidth or over a wide range of RF signal scenarios (i.e. what other devices are operating in the nearby spectrum neighbourhood) will be more complex and expensive than a single purpose device. One of the things that regulated spectrum gives you is a degree of certainty of your operating RF environment.

Cognitive radio is even more far out. The idea here is that some type of frequency division multiple access is used in the common spectrum and the individual radio “listens” and looks for unused spectrum to use. However, this requires some underlying common protocol that all the devices should use. Such protocols would have to accommodate both point-to-multipoint and mesh approaches. The alternative approach is a totally uncoordinated one where the only constraint that allows mutual co-existance is a power limitation (e.g. the WiFi model). CR algorithms are currently a hot topic in the academic research space but are quite far from being ready for implementation in the consumer space.

2) Regulatory and Standards delay: As I mentioned, CR will require some degree of regulatory and standards context. The standards context will define how CRs behave in terms of spectrum sensing, management, sharing and mobility. CR may not require detailed protocol definition but will certainly require some degree of specification, which will entail the involvement of bodies like the IEEE, ETSI, etc. And there would obviously need to be a corresponding change in the regulatory framework.

In general, the concept of spectrum commons is intuitively appealing. Unlicensed spectrum has already proven its value with the proliferation of WiFi and the spectrum commons approach dangles the possibility of extending the promise of unlicensed spectrum to a near utopian degree. This is always presented as an superior alternative to the sclerotic bureaucracy of the FCC making decisions on spectrum use. However, in the real world where people are actually building modems, radios and consumer devices, the regulatory context of the FCC provides more than just an economic model of how spectrum is used (i.e. spectrum as property with markets vs unlicensed or common spectrum). It also provides a technical context for engineers who design and build the technology. RF is pretty wacky stuff and although increasing computational power and antenna technologies are of critical importance and key enablers to new wireless architectures and protocols, they don’t eliminate the world of cavity filters, intermodulation distortion, adjacent channel interference, etc.

Ultimately, the either/or approach is problematic. Spectrum commons, like unlicensed spectrum before it, hold great promise and regulatory bodies should embrace it by making spectrum available. But it’s also 10 or 20 years away from being ready for primetime. There’s a lot of usable radio spectrum. The real answer is to embrace and enable multiple approaches and philosophies of spectrum usage.

Thanks for setting me straight. What I meant by solving the last mile problem directly was using a WiFi connection as a direct replacement for the local loop. In my understanding (correct me if I’m wrong) the range of WiFi is inadequate to do this in most cases: WiFi only reaches a couple of hundred feet, and many homes are considerably further from the local exchange.

I was not aware of the existence of CuWin, and I will check that out. I would have found Benkler’s argument a lot more persuasive if he had included a discussion of such networks, although maybe they didn’t exist yet when he began writing. Still, it looks to me like it’s still in the early testing phases. Their FAQ suggests that joining the network requires purchasing and configuring specialized equipment, something that most users are not going to do, and it states that “In its current state, it is best used for light surfing and not for heavy usage.”

Certainly, one reason for this might be that it doesn’t have adequate spectrum, and I’m certainly sympathetic to allocating additional spectrum to be allocated for use as a commons. But I think it’s still quite a stretch to cite the existence of a couple of small (and very expensive) proof-of-concept networks as a basis to argue, as Benkler does, that privately owned spectrum is now anachronistic. It’s worth experimenting with both approaches, but given the clear demand for privately owned spectrum (and the desirability of greater competition in the wireless market) it makes sense to continue allocating most of the spectrum that becomes available to exclusive ownership.

donut: Thanks for commenting. That’s very interesting!

“it had been rendered obsolete by technology. In particular, it had been rendered obsolete by the fact that the declining cost of computation and the increasing sophistication of communications protocols among end-user devices in a network made possible new, sharing-based solutions to the problem of how to allow users to communicate without wires”

The reality is that there are two key problems/challenges that will push out a spectrum commons approach for years (or decades):

1) The required technologies are far from being ready. Software defined radio (SDR) and cognitive radio (CR) are the two key underlying enabling technologies and both of these are far from being consumer technologies. SDR is used to build radios that support multiple interface technologies (e.g. CDMA, GSM and WiFi) with a single modem by reconfiguring it in software. However, SDR modems are expensive since they typically entail programmable devices like FPGAs, as opposed to the mass produced, single purpose ASICs that are used in most consumer devices today (and are key enablers for low cost handsets). Even today’s multi-mode devices tend to just have multiple ASICs (or multiple cores on a single ASIC). SDR is currently mostly used in military applications where cost is no object. SDR is also a modem technology and it ignores RF design issues. In particular, the RF design of a wireless device is typically closely coupled with the underlying access technology and modem design. E.g. different air interface technologies have different spectral mask requirements and different degrees of vulnerability to co-channel interference and strong adjacent channel power. A device that must work over a wide bandwidth or over a wide range of RF signal scenarios (i.e. what other devices are operating in the nearby spectrum neighbourhood) will be more complex and expensive than a single purpose device. One of the things that regulated spectrum gives you is a degree of certainty of your operating RF environment.

Cognitive radio is even more far out. The idea here is that some type of frequency division multiple access is used in the common spectrum and the individual radio “listens” and looks for unused spectrum to use. However, this requires some underlying common protocol that all the devices should use. Such protocols would have to accommodate both point-to-multipoint and mesh approaches. The alternative approach is a totally uncoordinated one where the only constraint that allows mutual co-existance is a power limitation (e.g. the WiFi model). CR algorithms are currently a hot topic in the academic research space but are quite far from being ready for implementation in the consumer space.

2) Regulatory and Standards delay: As I mentioned, CR will require some degree of regulatory and standards context. The standards context will define how CRs behave in terms of spectrum sensing, management, sharing and mobility. CR may not require detailed protocol definition but will certainly require some degree of specification, which will entail the involvement of bodies like the IEEE, ETSI, etc. And there would obviously need to be a corresponding change in the regulatory framework.

In general, the concept of spectrum commons is intuitively appealing. Unlicensed spectrum has already proven its value with the proliferation of WiFi and the spectrum commons approach dangles the possibility of extending the promise of unlicensed spectrum to a near utopian degree. This is always presented as an superior alternative to the sclerotic bureaucracy of the FCC making decisions on spectrum use. However, in the real world where people are actually building modems, radios and consumer devices, the regulatory context of the FCC provides more than just an economic model of how spectrum is used (i.e. spectrum as property with markets vs unlicensed or common spectrum). It also provides a technical context for engineers who design and build the technology. RF is pretty wacky stuff and although increasing computational power and antenna technologies are of critical importance and key enablers to new wireless architectures and protocols, they don’t eliminate the world of cavity filters, intermodulation distortion, adjacent channel interference, etc.

Ultimately, the either/or approach is problematic. Spectrum commons, like unlicensed spectrum before it, hold great promise and regulatory bodies should embrace it by making spectrum available. But it’s also 10 or 20 years away from being ready for primetime. There’s a lot of usable radio spectrum. The real answer is to embrace and enable multiple approaches and philosophies of spectrum usage.

By and large though, this is a limitation of the spectrum itself. Wifi is junk spectrum. Municipal wireless networks would be much different if spectrum with better properties (the ability to reliably pass through external walls, for example) was available as commons. The innovation (and competition) in the Wifi space is pretty compelling case for making more spectrum available as commons.

But on your actual point, a careful reading of Benkler reveals that he doesn’t simply have his own conclusions, he has his own facts. He insists, for example, that Wikipedia just as accurate as Britannica, and bases that on the Nature article. Not only did that article not say what Benkler thinks it says, even a casual reading of Wikipedia shows that it’s not true. Wikipedia is a nice place to go for low-value information, but that’s all it is, not an alternative source of high-value information.

Similarly, thanks to advances in coding (OFDM’s progeny) we may one day know how to build radios that can use the same (or nearly the same) frequencies as their close neighbors without interference, but that’s all we can say with certainty because the details are still very speculative. So it’s fine to speculate about the direction that spectrum policy might take if such a thing were reality, it’s not honest to assert that we have this super-duper radio coding technology today.

Wishful thinking seems to be at the heart of Benkler’s program.