Spectrum analysts (including me) often proclaim that harmful interference is a growing problem, or at the very least a growing risk. That sounds obvious, given the growing profusion of radios, packed more and more densely together in frequency and space. But could the opposite be true?
Data on interference trends are hard to come by. Published information covers just slivers of time, space and frequency. The Federal Communications Commission (FCC) doesn’t publish comprehensive data on interference incidents, in spite of repeated calls to do so. It does publish a non-comprehensive list of Amateur Radio Service enforcement actions, but that covers only 2009–2016. (The number of RFI incidents reported by amateurs is flat over this period.) When the FCC’s technical advisory council (the TAC) launched a formal inquiry in 2016 to find out whether there was an increase in background radio noise, some pointed to data sets but nobody offered a thorough analysis. The TAC didn’t come to any firm conclusions about whether radio noise was increasing or not.
Regulators could probably be more aggressive about cheek-by-jowl sharing arrangements, and more skeptical about incumbents’ claims that new arrangements will cause catastrophe
(We may have to live with poor data, but that shouldn’t stop efforts to improve measurement, and especially to start collecting long time series. Modelling past technology transitions, such as the conversion to digital broadcasting, communications and radar systems, could also help.)
It certainly sounds as if harmful interference is increasing, but our perception is skewed by cognitive biases. There is certainly selection bias: the people who report increases in interference are usually the ones that believe that such a rise exists. Availability bias also plays a role: since we tend to over-estimate the likelihood of events which are recent, unusual or emotionally charged, striking stories about exceptional interference sway our judgment. And last but not least, there’s confirmation bias: once we’ve taken a position – that harmful interference is getting worse, for example – we’re likely to search out, focus on, remember and interpret data in a way that confirms our preconceptions.
In the light of the absence of good data and concerns about cognitive bias, the most obvious null hypothesis is that the incidence of harmful interference isn’t changing. However, experience with other technologies suggests that the most plausible hypothesis is that it is, in fact, declining.
For example, consider car accidents in the United States. The US population more than doubled from 127 million in 1935 to 321 million in 2015, and the number of vehicle miles travelled (VMT) increased more than tenfold from 229 billion to 3,095 billion per year. However, the total number of road deaths stayed flat (34,494 in 1935 and 35,485 in 2015), and the number of fatalities per mile driven decreased tenfold from 15.09 to 1.15 per 100 million VMT*. There are similar trends in other technologies, like aircraft safety and air pollution.
There are many reasons for this. Advances in automobile technology such as airbags, crumple zones and anti-lock brakes played a major role in reducing fatal collisions – that is, in reducing “harmful interference” between vehicles. Stricter vehicle and occupant safety regulations and better enforcement were important too.
Radio technology has also improved dramatically, and those improvements have made radios less vulnerable to interference. Examples include transmitters leaking less interference outside their transmit bands, receivers becoming more resistant to interference, and more-directional antennas that only accept signals from desired directions. Hence, it’s reasonable to expect that there is less harmful interference today than there was in the past.
Regulation – such as requiring seat belts, and better enforcement of drunk-driving laws – also helped to reduce automobile fatalities. By comparison, there have been few new regulatory mandates to improve interference-rejection by radios. (The EU Radio Equipment Directive’s receiver requirements are the exception that proves the rule.)
Limits on transmitter leakage have hardly changed, and there are effectively no requirements on receiver performance. I don’t believe government-mandated receiver standards are the solution, though industry standards encouraged by government, such the new ETSI receiver standards, may improve matters. I prefer technology neutral ways to incentivize receiver performance, like harm claim thresholds; one automotive comparison is the U.S. Corporate Average Fuel Economy (CAFE) standards, where the government sets targets for average fleet fuel economy without telling manufacturers how to meet them.
So what if harmful interference is going down?
First, we might worry less about predictions of a spectrum crisis. Technology improvements seem to have allowed enormous radio densification without coexistence crises. Regulators could probably be more aggressive about cheek-by-jowl sharing arrangements, and more skeptical about incumbents’ claims that new arrangements will cause catastrophe.
Second, the success of the wireless community in avoiding a spectrum meltdown without stringent harm-avoidance rules and enforcement on both device performance and user behavior – such as in the automotive sector – suggests that light-touch regulation may be sufficient for wireless. (Whether this is indeed so, and why, is an interesting research problem.)
Third, even if the radio noise floor is going up, the risk of harmful interference may be constant or going down because systems are more interference tolerant. The glass-half-full implication is that we needn’t worry about the noise floor. The glass-half-empty interpretation is that our technologies are hiding a growing problem that may suddenly emerge with irreversible force, like market crashes, the eutrophication of lakes, and other tipping points in complex systems.•