Your browser is not supported. Please update it.

13 February 2006

IBM takes first step towards commercially viable 60GHz media network

Most of the attention of the players looking to bring fast wireless networks into the home is focused on low reaches of the spectrum that support long ranges and non-line of sight. However, there is also increasing interest in the higher frequencies, namely 60GHz, a portion of the so-called ‘millimeter wave’ spectrum that is license-exempt in many areas. IBM has been one of the leading players holding out the prospect of running UltraWideBand-style personal area networks in frequencies above 60GHz and last week it showed off its first developments in that direction, claiming it is heading quickly towards commercially viable implementations for the consumer market.

IBM stood up at the Solid State Circuits Conference in San Francisco and said it had a component that would allow electronic devices to transmit and receive 10 times faster than today’s Wi-Fi networks, with early prototypes operating at 700Mbps, and supporting the proposed 802.15.3c IEEE standard for 60GHz personal area networks. Really this was about using a combination of silicon and germanium to make radio chips that operate up in the rarified atmosphere of 60GHz, a line of sight range that has little wall penetration and which is currently used almost exclusively to create expensive line of sight wireless fiber, for high power, long distance applications, mainly mesh backhaul. The challenge, taken up by IBM, Intel, France Telecom and the IEEE standards body, is to create a low cost device for 60GHz that could take advantage of the high performance potential of the band, while getting round its obvious problems ‘ short range and the need for line of sight (to the extent that, in the home, a signal could be cut by someone walking round the room).

The breakthrough here for IBM is that previous chip designs trying to exploit this spectrum have been too large, expensive and difficult to build, often based on gallium arsenide, which is corrosive and easily infected by impurities. These technologies are required because of their extremely fast switching speed compared with simple silicon.

IBM says its design allows embedding of antennas directly within the chipset package, further reducing system cost and a prototype chipset module, including the receiver, the transmitter, and two antennas, would occupy the area of a small coin.

IBM is the global leader in chip packaging and a constant focus of its chips designs over recent years has been to reduce inter-chip communication, rather than directly drive up chip processing power. The company said that the technology could be targeted as a 60GHz wireless personal area networks in the 10 meters and below range. This is likely to be in extremely low power ranges that would cause no interference outside, say, a living room.

This technology might enable broadband video distribution to stream an uncompressed high definition video signal from a DVD player to a plasma display mounted on a wall. The trouble is that UWB technologies, particularly those from the WiMedia Alliance, are well positioned to take this market over the next 18 months or so, with the added advantage that they could operate in spectrum where the signal is not so easily disrupted. There are, however, many suggestions that signals such as WiMedia and WiMAX could be run over a 60GHz physical layer and the IBM process might feasibly be used as a way to build UWB chips, but even silicon germanium is unlikely to reach the price point that these chips will eventually need given that they will be manufactured in hundreds or millions once they are standardized. This makes efforts by Intel and others to massage CMOS techniques into this speed range all the more important.

The interest in 60GHz has been stimulated by the FCC decision in 2004 to open it up without licensing. Millimeter wave or ‘pencil beam’ signals, which operate in these high bands, are very narrow (10mm at 30GHz to 1mm at 300Ghz) and direct, which minimizes interference, and can support very high data rates over short point-to-point ranges. The opening up of this spectrum was lobbied for energetically by Intel, Microsoft and Cisco, all of which see its potential to support new applications and so spur uptake of wireless devices. In particular, Microsoft has focused on the ability to transfer a full length movie in about 10 seconds, potentially seeding video on demand and television services over WiMAX, Wi-Fi or cellular clients, or even ‘ chip advances permitting ‘ directly to the end user.

Two developments will be key if this vision is to become a reality. First, to create the standards that are essential to achieve the cost efficiencies and economies of scale to attract the big consumer electronics players; and to ensure these standards work with existing efforts, rather than obstructing them. Second, to develop chip technology that can be the basis of cost effective devices, a move in which IBM has made the early move.

Increasing congestion in 2.4GHz and risk of the same in 5GHz is making it important that other sources of unlicensed spectrum are exploited, especially if these are available on an international basis, as 57-64GHz is in many countries (in contrast with UWB-based solutions, currently only permitted in the US and, soon, China). Such bands will also be attractive since they tend to carry more generous power allowances than others ‘ the FCC’s limit in 60GHz is 40dBm, far higher than it permits for UWB in its current US allocated spectrum range of 3.1GHz to 10GHz, where its power limits are one-10,000th those of a cellphone .

Hence the interest of the IEEE, whose 802.15.3 groups address high data rate wireless personal area networks, increasingly with a view to setting standards for digital media networks. The main recent work has been done by the 802.15.3a taskgroup, which focuses on a UWB-based implementation, but this effort has now ended in stalemate and so the IEEE may be looking for an alternative route. Its 802.15.3c taskgroup is developing a physical layer standard for the 60GHz band which could be coupled with the 802.15.3 media access control (MAC) layer to deliver target data rates of up to 2Gbps. The group will review proposals later this year but a standard would be unlikely to be ratified until late 2007.

The application of the 802.15.3 MAC to multiple physical layers is a powerful way to create standards that are adaptable to changing spectrum allocations round the world, simplifying interoperability between systems in different bands and streamlining the development effort, avoiding the need to create an entire new standard for every frequency band. This will be important as regulators round the world start to respond to the need to open up more spectrum, and to be more flexible over permitted usage.

The ‘c’ effort may be a fallback option for the IEEE now that the 802.15-3a effort has failed, but this does not mean that the UWB technology is compromised, only the IEEE standard. Indeed, it is more likely that the WiMedia control layer, the frontrunner in UWB technologies, will be applied to 60GHz physical layers too in future. A logical extension of all this activity would be to use the UWB physical layer ‘ whether WiMedia or another ‘ in 60GHz too, leading to even greater harmonization and economies of scale.

Two issues make this outcome uncertain ‘ whether key regulators will extend the bands in which UWB is allowed to operate above 10GHz, and whether efficient chipsets could be designed. Some companies are already working on such projects though, although many are in the military space where UWB is allowed freer rein to span its whole spectrum range. Swiss components maker Huber+Suhner is one company with an R&D project in this area, looking at RF subsystems to combine 60GHz and UWB. Perhaps the most important research has been done by France Telecom R&D under the auspices of the EU IST-Magnet project examining personal area networks. France Telecom and its partners are examining how to implement UWB in 60GHz in order to benefit from the small wavelength and limited interference, hence permitting a higher radiated power level than classical UWB systems. Within the IEEE taskgroup, companies such as Philips, driven by the need to boost data rates for media networks, are prominent.

The main interest, from a chipmaker’s point of view, is that high frequencies support very small circuits and antennas. To some extent, these can be designed to compensate for the propagation problems. For instance, tiny antennas ‘ in the more futuristic projects, small enough to be part of a ‘body area network’ ‘ can still achieve high direction gain, or complex antenna arrays can be assembled to boost data rate and range even within the limitations of the frequency, using techniques such as beam forming or MIMO (Multiple In Multiple Out). The higher power limits allowed by the FCC also help make up for the inherent problems of 60GHz radios.

A project at Berkeley University in California is aiming to implement very small circuits in CMOS with a view to creating low cost, mass market miniature devices that will, in their first iteration, perform at 1Gbps over 10 meters ‘ a target the team aims to hit by year end. IBM is looking to build its radio initially in silicon germanium, believing that CMOS will only be appropriate once the market is established and starting to achieve volume. It will have evaluation hardware this year.

Clearly such projects will be of critical interest to Intel, whose ambitious designs on controlling the wireless chip market center on all-CMOS radios supporting one or multiple frequencies. It has several advanced projects in the area of CMOS radios for frequencies above 10GHz, and interestingly, seeks to establish the technology
potential for the full 802.16 standard. While WiMAX has been focused on spectrum below 6GHz, the 802.16 standard can be applied to higher spectrum up to 60GHz. If Intel can create an efficient CMOS radio for 60GHz, delivering advanced performance through MIMO and other techniques, we can be sure interest in applying WiMAX to those devices will be reawakened, both for reduced cost ‘wireless fiber’ backhaul and for multimedia distribution applications.