Print this article | Return to Article | Return to CFO.com
Four disruptive technologies are emerging that promise to render not only the next wave of so-called 3G wireless networks irrelevant, but possibly even their 4G successors.
Economist Staff, The Economist
June 21, 2002
It is more than a century since Guglielmo Marconi pioneered wireless data transmission. Yet, if the current pace of innovation in the field is anything to go by, wireless technology is still in its infancy. The surge in popularity of mobile phones — their number will overtake that of fixed phones during 2002 — has prompted both established firms and start-ups to investigate ways to make phones more efficient and versatile. At the same time, the Internet is going wireless, driving a separate wave of innovation as the Internet's legendary ability to disrupt traditional ways of doing things enters a new arena.
It is too early to see where all this might be leading, or even how these two waves of wireless enthusiasm will fit together. But the parlous state of the wireless-telecoms industry, and the difficulties surrounding the deployment of "third generation" (3G) networks in particular, could be taken as evidence that existing ways of doing things are reaching their limits, and that some radical new ideas are needed.
Here, then, are four emerging technologies that show much promise: smart antennas, mesh networks, ad hoc architectures, and ultra-wideband transmission. Smart antennas are already in use and mesh networks are starting to appear, while ad hoc architectures and ultra-wideband are still largely restricted to the laboratory. But each challenges existing ways of doing things; each, on its own, or in combination with others, could shake up the wireless world.
Wireless antennas, in their simplest form, are dumb. A base-station on a cellular telephony network, for example, typically communicates with nearby handsets by broadcasting in all directions, which is rather like shouting to ensure that everyone in a small room can hear you. Base-stations use only a fraction of the radio spectrum available, to avoid interference with adjacent cells, but the use of directional antennas enables radio frequencies to be reused more efficiently, thus boosting capacity. So instead of one omni-directional antenna, many base-stations now use three-directional antennas pointing in different directions, each of which covers a 120° sector.
Smart antenna systems go a step further, using multiple antennas to provide more accurate directional targeting and additional improvements in efficiency. "The base-station works out where you are from the relative signal strengths at multiple antennae," explains Marc Goldburg of ArrayComm, a smart-antenna firm based in San Jose, California. "Then it can direct its transmission." ArrayComm's IntelliCell technology is now deployed in nearly 100,000 base-stations in Japan, China and Taiwan. Adding IntelliCell technology to a base-station typically boosts capacity by a factor of three to seven, says Mr Goldburg. Metawave, another smart-antenna firm that is based in Redmond, Washington, claims similar benefits for its SmartCell "cell sculpting" technology.
In theory, says François Chin, a researcher at the Institute for Communications Research in Singapore, the capacity enhancement is proportional to the number of antennas, but in practice it is only possible to achieve about three-quarters of this improvement. His systems, which are being tested in Tokyo by NTT DoCoMo, Japan's leading wireless operator, typically achieve a five-fold capacity improvement with eight antennas. It is thus possible to support either more users, or the same number of users at a higher data rate, or to reduce the number of base-stations needed to provide a particular level of service.
The plunging cost of processing power means that smart antennas, which started out as an expensive military technology, are now a cheaper way to increase network capacity than building new base-stations. Yet the response of big infrastructure suppliers such as Nokia, Ericsson, Lucent and Nortel has been lukewarm. Nokia says there is no need for smart-antenna technology, because the price of base-stations is also falling fast. Ericsson, Lucent and Nortel have their own versions of the technology, but are not pushing it to operators. Marty Cooper, ArrayComm's chief executive, who pioneered cellular telephony at Motorola, smells a rat. "Manufacturers determine what technologies are used — and they want to sell more base-stations," he says.
But smart-antenna firms hope that the advent of 3G will work in their favour, since 3G networks will require far more base-stations than existing 2G networks, in order to provide extra capacity for graphics, video and other new services. The option to provide some of this extra capacity via smart antennas, rather than additional base-stations, ought to appeal to operators. "3G has driven the need for our technology to the surface," says Dr Cooper. He hopes that operators will put pressure on infrastructure suppliers to offer base-stations with smart antennas, and expects the technology to be ubiquitous by 2010.
In the meantime, ArrayComm is pushing ahead with a technology of its own, called i-Burst. It claims that i-Burst offers far better performance than 3G networks at a fraction of the cost. Retro-fitting smart antennas on to cellular networks can go only so far, says Mr Goldburg, because cellular networks were not designed with smart antennas in mind. By contrast, i-Burst is a mobile wireless-data technology that was designed specifically to exploit smart antennas. Compared with 3G, Mr Goldburg reckons it is about 40 times more efficient.
The fact is that i-Burst base-stations — equipped with smart antennas and colocated with the base stations of a 2G network — can provide a throughput of one megabit (1m bits) per second, at about one-thirtieth of the cost of building a 3G network for the same area. Thus, i-Burst plugs the gap between 3G networks (long-range, but capable of 384 kilobits per second) and the popular WI-FI wireless-network standard used to connect laptops to the Internet (short-range, and capable of 11 megabits per second). At the moment, the i-Burst receiver is a brick-like object about the size of a video cassette that fits on to the lid of a laptop, and draws about as much power as a WI-FIPC-card. But ArrayComm's licensees, including Kyocera, a Japanese electronics conglomerate, expect to be able to produce i-Burst receivers as PC-cards.
ArrayComm is pushing i-Burst in three separate ways. The company has bought spectrum in Australia and plans to launch a commercial service there in 2003. Hanaro Telecom of South Korea plans to launch i-Burst as a wireless broadband service this summer. ArrayComm is also trying to persuade firms that operate cellular-network towers on behalf of wireless operators to install i-Burst equipment. In short, Dr Cooper hopes that i-Burst will cut two technological Gordian knots at the same time, providing fast mobile-data access, and also a wireless solution to the "last mile" problem of providing high-speed broadband access to the home.
A Relay Race
Proponents of mesh networks also believe that they have found a way around the last-mile problem. At the moment, there are two main ways to provide broadband connections to the home: use either the local cable-TV network or a digital subscriber-line (DSL) from the local telephone company. DSL supercharges ordinary phone lines to enable them to carry data at high speed.
But not every neighbourhood has cable access, and DSL works only for subscribers close to a telephone exchange. Worse, the roll-out of broadband has been held back by obstructive telecoms incumbents, regulatory obstacles and infighting. No wonder the idea of a fixed wireless broadband service, blanketing a neighbourhood with connectivity without the need to lay any cables, seems so seductive.
The mesh-networking approach, which is being pursued by several firms, does this in a particularly clever way. First, the neighbourhood is "seeded" by the installation of a "neighbourhood access point" (NAP) — a radio base-station connected to the Internet via a high-speed connection. Homes and offices within range of this NAP install antennas of their own, enabling them to access the Internet at high speed.
Then comes the clever part. Each of those homes and offices can also act as a relay for other homes and offices beyond the range of the original NAP. As the mesh grows, each node communicates only with its neighbours, which pass Internet traffic back and forth from the NAP. It is thus possible to cover a large area quickly and cheaply. For providing fixed-wireless access, the mesh approach is technically superior to the traditional "point-to-multipoint" radio approach in a number of ways. For one thing, it requires much less power. Rather than using high power to get around obstacles, mesh networks offer multiple paths from one node to another; with systems typically being self-configuring so that, like the Internet, traffic is sent by the quickest route. Also like the Internet, mesh networks are robust and can be scaled up easily.
Another drawback of point-to-multipoint systems, observes Dave Beyer of Nokia's wireless-routers division, is their need for tall antennas to get above the clutter and maximise their coverage. Unfortunately, they then run into the problem of interference with adjacent cells. Mesh networks, in contrast, can use rooftop antennas, since each node needs only to be able to communicate with its neighbours. Such systems use one-ten-thousandth of the transmission power. That, in turn, means they can use unlicensed spectrum.
A number of firms are now pushing mesh-network technology as the fastest and easiest way to provide broadband Internet access. Following a successful trial in Santa Rosa, California, Nokia's system, called RoofTop, is being rolled out by more than 50 operators, mainly small Internet service-providers (ISPs). The ISP installs an AirHead unit (Nokia's name for a NAP) to seed a neighbourhood, and a small, weatherproof pod with an omni-directional antenna is fixed to the outside of each subscriber's home or office. Each pod costs around $800 — less if produced in large quantities. Vista Broadband, which offers a broadband service using RoofTop technology in Santa Rosa, charges around $200 for installation, and then a monthly fee of $50.
SkyPilot, a mesh-networks start-up based in Menlo Park, California, is taking a similar approach. Its rooftop units use smart antennas to beam data back and forth, enabling frequencies to be reused more efficiently and increasing capacity. Duncan Davidson, the firm's boss, says the Internet/mesh approach has many advantages over the traditional circuit-based approach used in telephony. "The Internet architecture gets better with density [whereas] the phone system gets worse," he says.
The problem with the mesh approach, however, is how to get it off the ground. Who will build the NAPs to seed a neighbourhood? Unlike Nokia, which simply sells its RoofTop gear to ISPs, SkyPilot plans to help prime the pump itself, by setting up NAPs and allowing ISPs to resell access. This approach also has technical merit: multiple overlapping mesh networks are far less efficient than a network in which all the nodes can talk to each other. So it makes sense to have competition at the ISP level, rather than the infrastructure level.
Perhaps the most ambitious vision of mesh networking is that of MeshNetworks, a firm based in Maitland, Florida. It has developed its own radio hardware and some clever routing software that makes it possible to blanket an area with broadband wireless coverage using "intelligent access points" (its term for NAPs) and shoebox-sized wireless routers. But what is really clever is that this wireless mesh-network then supports mobile devices, such as handheld computers and laptops. And those devices can also act as routers for other mobile devices, further extending the mesh. Cleverest of all, even when two or more devices are beyond the range of a NAP or a wireless router, they spontaneously form their own local network. MeshNetworks' technology thus combines the mesh architecture with the even more radical approach of "ad hoc" networking.
From the Battlefield
As the name suggests, ad hoc networks consist of multiple devices, each of which also acts as a router for the others. Furthermore, these devices may also be moving, so that the network topology is in constant flux. This poses a number of challenges, not least in routing. Clearly, the quickest way to send a packet of data from one device to another changes as the devices move around, and other devices join and leave the network..
Ad hoc networks are commonly associated with military and emergency applications, both of which have to operate in situations where there is no network infrastructure. For that reason, ad hoc networks are sometimes referred to as "infrastructureless" network architectures. Rescue workers in an earthquake zone, for example, could use handheld radios, each of which also acts as a relay for other nearby radios. Similarly, the robust, self-healing properties of ad hoc networks make them suitable for military use, either by mobile combatants, or to connect up "smart dust" sensors that would be sprinkled across a battlefield from an aircraft.
For many years, says Zygmunt Haas, a researcher at Cornell University, most research into ad hoc networks focused on military applications. Recently, however, interest in the field has increased as its commercial possibilities have started to emerge. "Bluetooth", a short-range wireless protocol that enables mobile phones to talk to nearby handheld computers, printers and other phones, is a simple form of ad hoc networking, though it supports only single "hops" between individual devices.
The advent of WI-FI networking equipment has also provided a foothold. With the right software, it is possible to allow WI-FI-equipped laptops to act as relays for other nearby machines, letting packets make multiple hops from machine to machine to get to and from the Internet. Dave Johnson, an ad hoc researcher at Rice University in Houston, Texas, has built demonstration systems based on WI-FI devices in moving cars that do exactly this. Ad hoc networking might also expand the capabilities of mobile phones. People attending the "Burning Man" festival in the Nevada desert would then be able to call each other, even without any local infrastructure, suggests Charles Perkins, an ad hoc guru at Nokia's research centre in Mountain View, California. As well as working without any infrastructure, ad hoc-capable mobile phones would have other advantages. In a crowded environment, such as a sports arena, phones could pass traffic from other phones to base-stations in adjacent cells, thus boosting capacity. Calls between users within the arena could be handled locally, without loading the cellular network.
The ad hoc/cellular hybrid approach would also improve coverage at the edges of a cellular network, since users just outside the network's range would be able to "hop" their calls into the network via somebody else's phone; in the process, they would extend the effective size of the network, allowing still more distant users to "multihop" their way in. No wonder Nokia and other mobile-infrastructure manufacturers are keeping a close eye on ad hoc networking.
But there are still several problems to overcome. The first is a conflict of interest: do you really want somebody in another row of seats using your phone as a relay and draining your battery? The trade-off, says Dr Haas, is that the service quality improves for all, at the cost of handling each other's traffic. Some proposed ad hoc architectures, he says, include micro-payment schemes to ensure that everybody pulls their weight.
Another difficulty is agreeing on protocols; ad hoc will work only if devices are ubiquitous, and support an agreed standard. But different situations require different standards. This may require hybrid, adaptive protocols, where the network's behaviour adjusts depending on the circumstances.
The ad hoc approach is also favoured by proponents of ultra-wideband (UWB) transmission. UWB marks a radical departure from existing wireless technologies because, rather than transmitting and receiving on a particular radio frequency, it involves transmitting very short pulses on a wide range of frequencies simultaneously at low power. Such pulses, which are typically less than a billionth of a second long, pass unnoticed by conventional radio receivers, but can be detected by a UWB receiver. Information is encoded into streams of pulses, millions of which can be sent every second, by varying their polarity or their timing relative to an apparently random but pre-arranged schedule. (A slightly early pulse might signify a one, and a late pulse a zero.)
UWB has been struggling to establish itself for years. That is because its unconventional approach requires regulatory approval. But its fortunes received a massive boost in February 2002, when America's Federal Communications Commission (FCC) gave limited approval for UWB transmissions, despite the objections of air-traffic controllers and telecoms firms worried that they might interfere with their existing networks. Similar moves are expected to follow in Europe and Asia, says Jim Baker of Time Domain, a leading UWB firm based in Huntsville, Alabama.
The FCC ruling limits the range of UWB transmissions to about ten metres, although longer ranges may be allowed in future once the question of interference has been sorted out. However, UWB is capable of a data rate of at least 100 megabits per second over such distances. So the way is now clear for commercial UWB products to provide wireless links between, say, personal computers and camcorders or music-players. Work is well advanced on a standard to enable UWB devices and peripherals to locate and communicate with each other. Such is the interest in the technology that heavyweights such as Intel are now actively investigating it, in addition to UWB firms such as Time Domain, Pulse-Link and Xtreme Spectrum.
Cellonics, a Singapore-based firm, has an interesting twist on UWB. Inspired by the firing of nerve cells, it has developed a circuit that generates high-frequency bursts of short pulses in response to an input signal, and which lends itself to UWB encoding. Cellonics recently demonstrated a short-range UWB system capable of transmitting at 11.4 megabits per second. It expects the first commercial applications for UWB to be in wireless-networking equipment for homes and offices. When higher-power transmission is allowed, says Lye Hoeng Fai, the firm's boss, he expects UWB to appear in cellular systems.
Together, UWB and ad hoc architectures are a natural fit, since the UWB devices will have to locate each other and start communicating automatically, tasks that ad hoc networking readily facilitates. The two technologies are thought to have been used together in military applications for some time. UWB pulses, emitted apparently at random, are very difficult to detect or intercept, and are ideal for battlefield transmissions. UWB pulses can also be used for medical imaging, high-resolution radar, and proximity detection. But it is their potential use in communications that is arousing the most interest. Looking beyond 3G networks, and the patchwork of WI-FI and cellular networks that is often referred to as 4G, some are even referring to infrastructureless, ad hoc UWB networks as 5G.
Turned Upside Down
There is thus no shortage of new wireless technologies. But how these and other innovations will shape communications networks remains to be seen. On the one hand, there is a clear trend towards consolidation in wireless telecoms — with the likelihood being that there will be only a handful of global wireless operators by 2010. On the other, many emerging wireless technologies seem to signal a move to a less informal, more decentralised model along the lines of the Internet. Dr Johnson at Rice University suggests that ad hoc networking will create more scope for "mom and pop" network operators and free community networks, all stitched together in a casual, ever-shifting web.
Network operators will still be needed to carry long-haul traffic, but their role could become less (rather than more) important in future. In the process, the entire structure of the industry could shift from a top-down approach to one that is organised from the bottom up. There are already signs of this happening in the emerging area of commercial WI-FI networks, which allow individuals to club together to form a larger network. The business models and billing systems remain uncertain. But one thing is clear: it is still very early days for wireless data.