Shan't!

SCIENCE & TECHNOLOGY
Wireless broadband

Jan 6th 2005
From The Economist print edition

Computer chips for “open-spectrum” devices are a closed book

TELECOMMUNICATIONS used to be a closed game, from the copper and fibre that carried the messages, to the phones themselves. Now, openness reigns in the world of wires. Networks must interconnect with those of competitors, and users can plug in their own devices as they will. One result of this openness has been a lot of innovation.

Openness is coming to the wireless world, too. Cheap and powerful devices that use unlicensed and lightly regulated parts of the radio spectrum are proliferating. But there is a problem.
Though the spectrum is open, the microprocessor chips that drive the devices which use it are not. The interface information—the technical data needed to write software that would allow those chips to be used in novel ways—is normally kept secret by manufacturers. The result could be a lot less innovation in the open wireless world than in the open wired one.

Take, for example, the Champaign-Urbana Community Wireless Network (CUWiN), in Illinois. This group is trying to create a so-called meshed Wi-Fi network. Wi-Fi is a wireless technology that allows broadband internet communication over a range of about 50 metres. That range could, however, be extended if the devices in an area were configured to act as “platforms” that both receive and transmit signals. Messages would then hop from one platform to another until they got to their destination. That would allow such things as neighbourhood mobile-phone companies and a plethora of radio and TV stations, and all for almost no cost. But to make such goodies work, CUWiN needs to tweak the underlying capabilities of Wi-Fi chips in special ways.

When its engineers requested the interface information from the firms that furnish the chips, however, they were often rebuffed. A few companies with low-end, older technology supplied it. But Broadcom and Atheros, the two producers of the sophisticated chips that CUWiN needs if its system is to sing properly, refused. Nor is CUWiN alone in its enforced ignorance.

SeattleWireless and NYCwireless, among other groups, have similar ideas, but are similarly stymied. Christian Sandvig of the University of Illinois at Urbana-Champaign, who has been studying the brouhaha, believes regulators ought to enforce more openness.

Broadcom and Atheros say that making the interface information public would be illegal, because it could allow users to change the parameters of a chip in ways that violate the rules for using unlicensed spectrum (for example, by increasing its power or changing its operating frequency). That is a worry, but it depends on rather a conservative interpretation of the law.

The current rules apply to so-called “software-defined radios” (where the ability to send and receive signals is modifiable on the chip), and do not apply directly to Wi-Fi. Also, by supplying the data, manufacturers would not be held liable if a user chose to tweak the chip in unlawful ways. And in any case, if the firms are really worried, they could release most of the interface, keeping back those features that are legally sensitive.

Nor is the interface information commercially sensitive. Engineers are not asking for the computer code that drives the interfaces, merely for the means to talk to them. And having the interface information in the public domain should eventually result in more chips being sold. So it is hard to see what the problem is beyond a dog-in-the-mangerish desire not to give anything away. Time to open it up, boys.

Copyright © 2005 The Economist Newspaper and The Economist Group. All rights reserved.