SAN JOSE, Calif. In the race to a gigabit-per-second wireless link for the digital home, some ultrawideband backers are gasping for air while Wi-Fi proponents are revving up for the next lap.
An IEEE 802.11 study group is gearing up to launch a standards effort next year that could pave the way for Wi-Fi to step up to 3-5 Gbits/second data rates. Meanwhile an independent tester reports two UWB products now shipping have average throughput of just 20 Mbits/second at a range of 15 feet.
“We are finding throughput is quite disappointing,” said Fanny Mlinarsky, an industry expert in wireless test. “There is nothing above 50 Mbits/s maximum, and the average is 20 Mbits/s. Everyone thought this was going to be the short-range Gbit network,” she said.
Mlinarsky's lab, Octoscope, is conducting tests of shipping UWB systems sponsored by UWB vendor Pulse~Link Technologies. So far, companies shipping UWB silicon have declined to participate in the tests, many citing the fact a competitor is sponsoring the program. Jack Shandle, editor of sister Web site Wireless DesignLine, has publicly called on UWB chip makers to participate in the tests, so far without effect.
Vendors said inefficiencies in today's media access controller chips and software drivers are inhibiting throughput, something that should be cleared up in future products. Mlinarsky plans to acquire a LeCroy tester to check the physical layer performance of the systems.
“We are getting less than a tenth of the PHY data ratethat can't be all MACs and drivers,” she said.
Octoscope tested systems from Belkin and IO Gear using chips from Alereon, Freescale, Intel and NEC. It plans to test Toshiba systems using other silicon this week. Pulse~Link did not ship its test chips to Octoscope before fires swept through the San Diego area this week, cutting off some communications.
Mlinarsky believes the orthogonal frequency-division multiplexing technology used in the wireless USB version of UWB may be the culprit.
“Everyone joined Intel in the WiMedia Alliance before they did due diligence on the technology,” Mlinarsky said. “OFDM is not the optimal choice at these low power levels, so the industry may have made a mistake,” she added.
Companies doing their own tests of UWB silicon agree that performance of the parts is so far unacceptable.
“I haven't seen anything I want to build a product around. We can't afford to design in a 1W radio to get less than 100 Mbits/s over a few feet,” said a senior technology manager at a cellphone company who asked not to be named.
The UWB problems may be just a passing storm, said wireless analyst Craig Mathias of Farpoint Group (Ashland, Mass.). But the problems need to get addressed soon or other technologies such as advanced Wi-Fi or emerging 60 GHz radios will steal UWB's thunder, he added.
“We think UWB can get to 200 to 300 Mbits/s at the application layer without too much trouble,” said Mathias. “We think this is just a matter of early chips and drivers [but] UWB doesn't have this market all to themselves,” he added.
Indeed, last week IBM backed an emerging push to 60-GHz radios, an approach pioneered by startup SiBeam for a variety of consumer uses. IBM said it will make in its silicon germanium process radio that can be paired with baseband chips from MediaTek and other companies.
Another startup, New LANs, also is pursuing 60-GHz radios. Meanwhile other companies such as Amimon and Radiospire are rolling out their own approaches to Gbit-class wireless links for a range of consumer applications.
For its part, the IEEE 802.11 convened a Very High Throughput study group to look at Gbit-class versions of Wi-Fi. To date, the group has heard a range of presentations from AT&T, Intel, Motorola, Nokia and startup Wilocity. It could be ready to launch a formal standards effort as early as January.
One presentation mentioned the possibility of building 3 to 5 Gbit/s products in the 5-GHz band that could serve a range of uses including wireless links to monitors, projectors and video cameras. Those are some of the same apps UWB is attacking.
A presentation from AT&T Labs called on the group to raise its voice against spectrum allocation plans in Europe that could shut Wi-Fi out of the 275 GHz to 1 THz bands.
“The best terahertz spectrum is being carved up among science, satellite and amateur radio interests. The give-away of this precious spectrum is unexpected, capricious and irreconcilable with the evolution of future broadband wireless networks,” David Britz of AT&T said in his presentation.
Britz also noted that startups such as Phiar Corp. (Boulder, Colo.) are already working on terahertz-class transceivers. “The 2.4 and 5 GHz bands are getting crowded, and 60 GHz is looking like the promised land for high performance LANs,” added Mathias.
Meanwhile, the many startups pushing UWB could be running into trouble keeping investors interested. Several sources report Tzero Technologies, one of the early hot startups in UWB, may be taking a hard look at its options.
“We are in the midst of a funding round and its OK, but not going as rapidly as we would like,” said a company spokesman. “M&A is one of the possibilities. We have talked to strategic investors for awhile. It's anybody's guess how things will end up,” he added.
Tzero has working UWB chips that can deliver more than 200 Mbits/s at the applications layer, he said. But the company has no independent testers who can verify that claim and has so far opted out of the Octoscope tests.