A Snake in the Briar Patch: Radio Interference

One of the ongoing issues in spectrum policy is the question of “causing interference to others”. It was at the heart of the LightSquared controversy; it’s central to the spectrum sharing idea that the White House has been pushing since the publication of PCAST report on government spectrum; it’s a major obstacle in repurposing spectrum from many legacy uses to most modern ones; and it’s the central problem with Wi-Fi/LTE-U coexistence.

One foray in the effort to make sense of interference is a rather lengthy pair of blog posts by Public Knowledge’s Harold Feld. Feld is afraid that LTE-U will destroy Wi-Fi:

The 3.5 GHz band was a likely target for QCOMM and Friends to deploy their flavor of LTE-U. Not only was it going to become available, but it fit QCOMM’s general goal of screwing up the PCAST approach. Ever since the FCC opened the 3.5 GHz proceeding in 2012, QCOMM and friends have pushed to make it as much like regular licensed and as little like anything actually new or potentially disruptive to the Established Order as possible. But suddenly, at the end of 2014/beginning of 2015, QCOMM and Friends were all interested in getting 3.5 GHz moving and capable of supporting LTE-U.

This paranoid flourish is indicative of a bias that becomes evident when Feld makes his actual point about 13,900 words into the analysis:

LTE Dependents would still have to show enough evidence before deploying LTE-U that there is good reason to believe this isn’t going to blow everything up, but they would not have to remain trapped in permanent testing Hell or adopt a specific co-existence protocol like LBT. Nor would this be a safe harbor. In the event an unforeseen disaster starts to occur, the FCC can force LTE-U to back off and play nice, even if that means reducing power or altering the duty cycle. But LTE Dependents could go ahead, subject to preliminary testing that their coexistence claims work out. If they alter the coexistence mechanism later – especially if the alteration appears designed to provide an anti-competitive advantage over the Wi-Fi dependents – the FCC can act immediately to require the LTE Dependents to stop mucking with the co-existence mechanism. Since LTE Dependents know that they muck with the coexistence standard at their own risk, they cannot complain about being treated unfairly.

Do you see the problem?

Feld assumes that Wi-Fi systems have a property right to government spectrum in the 3.5 GHz band, and LTE-U systems can only use this spectrum if they can prove to the FCC that their transmissions don’t “blow Wi-Fi up”; I think that’s supposed to mean that they don’t cause “destructive interference”. That’s a policy assumption that obviously doesn’t stand up to any sort of scrutiny at all, so let’s simply look at the technical side of the question assuming that Wi-Fi has a property right.

If a person or company with an interest in using LTE-U in the 3.5 GHz band submitted a system to the FCC for testing, or more realistically, submitted a test report to the FCC proving an absence of “destructive interference”, what would that report need to show?

Right off the bat, the LTE-U is asked to prove a negative: they have to show that they don’t generate “destructive interference” regardless of how well designed or constructed the incumbent Wi-Fi system is. Leaving aside the fact that there aren’t incumbent Wi-Fi users in the 3.5 GHz band (try to make your Wi-Fi access point use it and you’ll find it’s not on the menu), what would the test procedure need to show?

If it runs a test where a Wi-Fi system transmits at a fixed rate, and then an LTE-U system joins the party at a fixed rate, would a slight reduction in Wi-Fi throughput be evidence of “destructive interference”, or would the Wi-Fi system need to stop altogether? How about a major but not catastrophic reduction in Wi-Fi throughput? This is unclear to say the least.

Evaluating Interference

And more meaningfully, how should the FCC judge the results if the this test case produces the same degradation in Wi-Fi performance as we would see with two pairs of Wi-Fi systems operating at the same time?

The issue, of course, is that every user of a network commons degrades every other user to the extent that the degraded user desires to use the entire commons at any particular time. This is the problem of contention for a shared resource, and it only goes away if the two systems are designed in such a way that neither interferes with the other. Non-interference could be accomplished for radios by making the beams so narrow that the energy from the LTE-U transmitter doesn’t fall on the Wi-Fi receiver, or vice versa, or that none of the Wi-Fi energy falls on the wrong receiver. In the millimeter wave spectrum (60 GHz and higher) this is possible, but it’s not entirely practical in 3.5 GHz today.

This leaves us with another design choice that can’t really be made at the transmitter side but can be done with varying degrees of success at the receiver side. Transmitters and receivers are designed to form and decode bits in particular ways and only in those ways. So LTE-U could be designed to form bits in such a way that Wi-Fi would see LTE-U bits as noise and filter them out.

But this would take some clever engineering on the part of LTE-U chip designers and a great deal of testing with a broad spectrum of Wi-Fi devices to assure success, pardon the pun. If the onus of compliance falls on LTE-U designers, Wi-Fi chip designers have no motivation for cooperating with the testing, let alone making design changes to assure co-existence.

And we know from Internet interconnection agreements that unbalanced negotiations don’t produce optimal results.

How does anyone prove a negative?

In the general sense, you can’t. All you can do is evaluate a finite number of test cases and report the results. If a Wi-Fi system fails to operate in the presence of LTE-U, or vice-versa, that can be the fault of either the transmitter or receiver design of either system. So if we’re looking for co-existence, the designers need to cooperate, and they’re not going to cooperate unless motivated.

I don’t believe Wi-Fi is interested in 3.5 GHz to begin with. There’s been an IEEE 802.11y standard for sharing spectrum with government for more than 10 years and nobody uses it.

So the entire discussion is actually moot. Sorry for wasting your time.