Uncharacteristically for the FCC, the technical reasoning does not stand up to close study.

The FCC has turned down a request to use a narrower bandwidth than the rules require.

Narrower bandwidth? No, it’s not a typo.

The decision turns on what may be the single most-used section in the entire five-volume FCC rulebook. The lawyers and engineers call it Section 15.247, but most people know it by the vast range of unlicensed products this rule has made possible: Wi-Fi, Bluetooth, modern cordless phones and nursery monitors, self-reading electric meters, ZigBee industrial equipment, wireless headphones and speakers, remote-control helicopters . . . the list is very long.

As originally adopted in 1985, Section 15.247 authorized “spread spectrum” devices. These, by design, disperse a signal over more bandwidth than the signal might otherwise require. The technique offers several advantages: less interference to other users than the equivalent narrowband signal; lower susceptibility to received interference; and encryption-like properties that make the signal difficult to intercept. Initially, Section 15.247 allowed two forms of modulation: “direct sequence” spread spectrum, similar to the CDMA modulation used by Verizon and Sprint for cell phone service, and frequency hopping, which is the basis for Bluetooth and many other technologies.

To enforce the “spreading” requirement, the FCC imposed a minimum bandwidth of 500 kHz, required the signal to be “deliberately widened” (the rules did not say by how much), and mandated a property called “processing gain,” which supposedly measured the device’s resistance to incoming interference. We add the “supposedly” because engineers in those days disagreed vehemently on whether the permitted tests for processing gain in fact measured that property, and indeed whether they measured anything useful at all.

In the meantime, demands for speed in wireless devices kept increasing.

By the late 1990s, the two megabits/second offered by Wi-Fi version “b,” permitted by the then-current Section 15.247, no longer kept pace with many consumers’ needs. But a company that tried to certify a device under what became Wi-Fi standard “g,” at up to 54 megabits/second, was turned down for not spreading its signal enough. The higher the data speed, the FCC reasoned, the more spectrum it should have to occupy. The company protested that it complied with the rule, vague though it was. The FCC upheld its decision, but in 2002 it changed the rule to let in the company’s technology. In particular, by dropping the requirements for spreading and processing gain, it made possible Wi-Fi g, and later, n. (See details on the differences here.) 

But the FCC kept the 500 kHz minimum bandwidth, by that time largely an historical artifact. The FCC logically could have allowed lower bandwidths at reduced power, as it effectively did for bandwidths over 500 kHz, where it caps the power at 6.3 milliwatts within any 3 kHz. But the 500 kHz minimum remained in place, for reasons the FCC never tried to explain.

Then came Starkey Laboratories, which manufactures hearing aids and related equipment. Starkey wanted to improve sound quality for people who wear hearing aids in classrooms, auditoriums, places of worship, and the like, by transmitting audio information under Section 15.247. Owing to technical limitations in hearing aids, Starkey wanted to use signals that occupy only 100 kHz. But it offered to reduce the power proportionately, to comply with the 6.3 milliwatts per 3 kHz limit. Procedurally, it asked for either a waiver or a rulemaking.

The FCC said no. The rulemaking Starkey requested “plainly does not warrant consideration,” it said, and to grant a waiver would “undermine the underlying purpose” of the rules.

Usually we agree with the FCC’s technical decisions. But this one frankly baffles us.

The FCC based its ruling on several findings: that Starkey failed to demonstrate its device would not increase the potential for interference to licensed services, and would not disrupt the “ecosystem” of multiple unlicensed devices in the same bands; and that a waiver would undo the FCC’s goal of barring narrow-band systems from using the relatively high power limits under Section 15.247.

But wait a minute. If Starkey had offered a one-watt device occupying 500 kHz, the FCC would have approved it without a second thought. Get out the calculator: that permissible one-watt device, if it uses spectrum uniformly, puts the same 6.3 milliwatts into each 3 kHz of bandwidth. (Ignore the small rounding error.) Starkey wanted to use a permissible level of power-per-spectrum, just over less spectrum than a compliant device. That creates less potential for interference, not more. And it does nothing to violate the prohibition on narrow-band devices using high power: Starkey would have to stay under 0.2 watts, rather than the full watt allowed to 500 kHz devices.

If the FCC were on the fence, we would think the public interest in assisting the hearing-impaired would have pushed it toward a yes.

The FCC’s decision may well be the right one. But if so, it must have relied on reasoning it did not spell out in the order turning down Starkey.