Interest has been rising for several years in using the 60GHz spectrum for ultra-high speed wireless communications, especially for in-home digital networks. This ‘millimeter wave’ band can support multi-Gbps signals, and in many countries is license exempt and in plentiful supply, so the potential for a mass market in home media is clear. However, there have been serious obstacles in the way ‘ notably the cost of the chips and the limitations, at high frequencies, on range and non-line of sight support. Now semiconductor companies are just starting to address these issues in order to position themselves for key standards body decisions from next year. The latest breakthrough comes from one of the first pioneers in commercial 60GHz solutions, IBM, which has teamed with Taiwan’s MediaTek to develop a chipset for wireless personal area networks, which the allies will submit to the IEEE for consideration for its proposed 802.15.3c WPAN standard.
The choice of MediaTek is significant as the company is one of the world’s largest fabless semiconductor designers geared to consumer mass markets. IBM, meanwhile, was one of the first companies to demonstrate working 60GHz prototypes suited to consumer gear, showing off a prototype back in February 2006 that was compact enough for home equipment and used silicon germanium (SiGe) rather than the corrosive and prohibitively expensive gallium arsenide (GaAs) employed in earlier 60GHz designs. IBM claimed to have used its advanced packaging skills to match the ultra-fast switching capabilities of GaAs in the usually slower SiGe, embedding the antennas directly within the chipset package to reduce size and power.
Working with MediaTek will help IBM adapt such breakthroughs for mass consumer markets, and the two companies are promising a chipset next year incorporating IBM’s 130nm BiCMOS SiGe radio and MediaTek’s baseband. IBM believes this will be a key technology to eliminate wires from the home, and could run a variety of MAC layers, including ones derived from UltraWideBand or even other IEEE standards like WiMAX. It also expects to work with other baseband makers in future, to start to create economies of scale and reduce costs once devices start to come to market near the end of the decade. IBM said its original GaAs prototype cost $10,000-$20,000 per radio; with BiCMOS, which implements the 60GHz transistors in SiGe and the rest in CMOS, the cost is reduced to $100-$200.
“We want to eliminate the jumble of wires consumers need now to interconnect all their different components, from the living room to the office,” said Mehmet Soyuer, lead researcher for Communication Technologies at IBM’s TJ Watson Research Center in Yorktown Heights. “Our radio provides a whole room with wireless high speed connections; depending on the modulation scheme, ranges of five to six meters are typical.” Since high frequencies support only short ranges, it is expected that, in a wireless home, 60GHz would be used to blanket a single room in 2Gbps-plus connectivity, while a longer range system like Wi-Fi or UWB-based WiMedia would work between rooms or floors (indoor WiMAX has even been proposed for this purpose).
The competition to take a dominant position in the development of this potentially significant technology will be fierce, and already the industry groupings are forming to try to impress the IEEE and the major consumer electronics vendors and alliances. An important breakthrough came earlier this year from Toshiba, which is a founder member of the WirelessHD group, a consortium of vendors developing 60GHz platforms and hoping to set de facto and industry standards. The Japanese giant succeeded in doing what even IBM has not yet demonstrated, and implementing a 60GHz chipset entirely in CMOS, promising even lower costs.
The company’s test chip measures 2.4mm x 1.1 mm, and all required receiver function blocks are integrated, including an antenna, low noise amplifier, mixer and PLL synthesizer, using a 90nm CMOS process. “We integrated a PLL synthesizer on chip and believe that it’s the first chip that operates alone as a receiver,” said Toshiya Mitomo, a researcher at Toshiba’s Mobile Communication Laboratory, presenting the device in Kyoto in June. One of the reasons for the expense and bulk of GaAs modules was that they required the antenna and synthesizer to be integrated separately using bonding wire. Toshiba says its advance will initially reduce fabrication costs to 10% of those of a GaAs alternative (similar to IBM’s promise for BiCMOS), and that further economies will be achieved in the medium term.
Toshiba’s key contribution has been to address the problem that, when CMOS process technology is applied at high frequencies, signals are easily leaked and chip operation can become unpredictable. To address this, Toshiba used a differential structure that is common in cellphones but rare in 60GHz products, and sets up noise and parasitic elements to, in effect, cancel one another out. The company still needs to create an amplifier before it can commercialize its chip, and has not yet announced the partners it will require to offer a full OEM solution like IBM’s.
WirelessHD was formed in November 2006 by six of the world’s biggest consumer electronics companies – Sony, Panasonic, Samsung, Toshiba, NEC and LG Electronics ‘ initially appearing to strike a blow to another fast WPan technology, UWB, which lies at the heart of various contenders for the digital home market, most notably the Intel-backed WiMedia. However, the future is most likely to revolve around integration of different technologies, something already heavily pushed by WiMedia, which will provide a next generation physical layer for multiple standards including Bluetooth and Wireless USB, as well as generic IP. So eventually, an integrated platform combining UWB, 60GHz, Wi-Fi and Bluetooth could also be on the cards to address all the needs of the media network.
The IEEE 802.15.3c task group accepted 16 proposals for its standard in May, heralding a likely new political battle like the ones that delayed 802.11n and derailed the UWB-based WPan activities. The WirelessHD supporters’ group is one candidate – although Sony and Toshiba are widely cited as lead creators of the specification, much of it was based on the work already done by 60GHz specialist SiBeam. Other contenders to direct the standard will include IBM, Philips and its former chip arm NXP, and another Japanese dominated group called CoMPA (Consortium of Millimeter Wave Practical Applications), which also submitted its 60GHz proposals in May and which claims to have the greatest market weight behind it, with 20 members, nearly all from Japan (some members belong to both, notably Sony and NEC). The CoMPA technology can reach a peak of 4.8Gbps over 10 meters in a frequency band ranging over 2,080MHz per channel. The companies expect to prototype a system, with or without IEEE acceptance, by the end of this year, targeting applications such as home high definition video transmission and video downloads from kiosks.
One of the key vote casters will be Intel, which beyond calling for 802.15.3c to be OFDM-based like WiMedia and 802.11n, has not submitted a proposal nor thrown its weight behind any of the groups. Given Intel’s market weight as a chipmaker, its ambitions in the CE world, and its power and expertise in working the IEEE processes, its support will be keenly sought by all parties. One route that Intel is likely to favor in future is convergence between different WPan technologies, particularly a combination of UWB with 60GHz spectrum. However, the giant has been ambivalent about 60GHz, despite doing R&D projects of its own on CMOS implementations. Stephen Wood, head of the WiMedia Alliance and an Intel employee, said last fall: “The regulations for 60GHz radios are not all in place, and the standards aren’t ready yet.” He claims the Alliance had studied 60GHz and concluded it would not be market ready for at least two years. The need for multiple sources of chips and interoperability testing would stretch time to market still further, he said.