FOR most people modern technology is a nightmare of impenetrable acronyms and standards. Does your phone have Bluetooth? Your laptop have Wi-Fi? Is your PDA GPRS-enabled? Does your car come with built-in GPS? Should you care?
As long as it works, perhaps not. Often the only reason you need to know which one of these a device supports is to know if it can communicate with another one. But that may soon change. On the way are intelligent devices that can autonomously choose the best radio signal, based on the local conditions around it and on how busy each frequency is. Prepare to be rescued from the electronic Tower of Babel.
This latest revolution is not just about neat gadgets for lay people. It will also force a radical rethink of who owns and controls the airwaves. To some experts it even paves the way for a complete overhaul of radio communications and instantly solves one of the biggest problems in radio communications: interference. 鈥淩adio interference doesn鈥檛 really exist,鈥� says David Reed, co-director of the viral communications lab at the MIT Media Laboratory. 鈥淚t is not a property of radio waves, but rather a by-product of badly designed receivers.鈥�
Advertisement
Behind this scientific and economic earthquake is an idea that has in fact been around for over a decade: software defined radio (SDR). Essentially it is just software that emulates the hard-wired functions of a traditional radio (快猫短视频, 17 May 1997, p 26). But only now have powerful computer chips become cheap enough to realise the promise of SDR on a grand scale. 鈥淭wenty years ago it was pointless to run SDR on anything but a supercomputer. Today we all carry around that much computing power in our pockets,鈥� says Reed.
The principal difference between SDR and standard radio is simple. When you wire together the components of a hardware radio transceiver (transmitter-receiver), they only respond to a particular frequency band and type of modulation (the variations in frequency or amplitude that let you piggyback information on the carrier wave). To change the sort of signal the radio can pick up 鈥� FM instead of AM, for example 鈥� you need to wire up a completely different circuit using different components.
SDR changes all that. You still need an antenna and an amplifier to receive and boost received radio waves, but these are not signal-specific (see Diagram). SDR then uses hardware called an analogue-to-digital converter, which constructs a digital approximation of the smooth radio signal. The software can then process this digital representation to search for any signal that might be there, with the only limiting factor being the size and shape of the antenna. And because software switches tasks so readily, you can change which sort of signal you are looking for practically in real time.
For example, a software-based mobile phone could switch from normal cellphone frequencies to the much faster Wi-Fi band whenever it is near a hotspot. It may sound far-fetched, but Intel has already produced such a device. Its prototype, dubbed the Universal Communicator, integrates four different communications schemes: GSM (ordinary cellphone networks), GPS, Wi-Fi and Bluetooth. It lets you place a call over the standard GSM network, and then seamlessly passes the call to a Wi-Fi hotspot if you come within range. The device uses GPS to pinpoint your location outdoors and then undetectably switch to a Wi-Fi-based positioning system indoors (快猫短视频, 26 June, p 21).
Intel鈥檚 prototype seems innocent enough. It used existing standards because they show something that can work immediately. But that belies SDR鈥檚 potential for disruption: you can broadcast anywhere on the spectrum in a blink of an eye and at virtually no extra cost. The implications are causing a storm.
In most countries today, the right to broadcast on the bulk of the radio spectrum is strictly licensed by regulators. For example, in the US the Federal Communications Commission (FCC) doles out licences to radio broadcasters, while the UK has a communications regulator called Ofcom. In each country the radio spectrum is divided up by frequency into unlicensed and licensed bands. For example, Wi-Fi operates on an unlicensed band around 2.4 gigahertz, whereas GSM, FM radio and UHF television broadcasts all use licensed frequencies. The licensing was put in place to ensure as little interfere as possible on each frequency, a pitfall easily demonstrated by switching on a hairdryer near a TV set 鈥� the ensuing fuzz on the screen is caused by radiation from the drier鈥檚 motor, which acts a bit like a random noise radio transmitter.
SDR is fundamentally challenging this old-fashioned approach to regulating the airwaves. 鈥淚t鈥檚 an accident of history,鈥� says Reed. 鈥淭he existing framework was chosen because when radio was invented all we knew how to do was filter out signals according to frequency.鈥�
As an analogy, imagine dropping two stones into a pond in different locations. As the ripples from the stones radiate and overlap, a pond skater sitting at some point on the pond鈥檚 surface would be aware that the surface was moving up and down, but it couldn鈥檛 tell there were two separate stones responsible for this wave. From its perspective the two signals have combined to form a third.
Similarly, if a single receiver happens to pick up two radio signals from different sources, it can鈥檛 distinguish the two: all it hears is the sum of both signals at that point. One signal is commonly said to cause interference with the other, yet as anyone who鈥檚 watched the ripples on a pond will know, when waves meet they hardly interfere at all, and actually pass through each other. The same happens with radio waves. The term interference is misleading, if not downright wrong, says Reed. 鈥淔undamentally, two different radiation sources don鈥檛 interact,鈥� he says. Although radio waves from separate sources might overlap, and a radio listening in might pick up a distorted transmission, the information in each signal is not altered or destroyed. 鈥淲hat we perceive as interference is in fact poor receiver design,鈥� he says. 鈥淵ou can always reconstruct the waves if you have enough information.鈥�
Each receiver hears a slightly different blend of signals depending on its location. The key to teasing them apart is having multiple receivers listening at different points and then pooling their information to separate out one signal from another. It鈥檚 a bit like having an overview of the radio signals in the same way we watch the surface of the pond. Once you can do that, then you can start transmitting a virtually limitless number of signals simultaneously at any frequency.
It sounds too good to be true, but Reed is already working on ways of making it happen. However, it will mean all radios will have to be able to transmit as well as receive. 鈥淲e鈥檝e grown up most of our lives with the idea that receivers exist without transmitters, but those radios are very handicapped in this sort of environment,鈥� says Reed. If he gets his way, the impact will be profound. 鈥淲hat we鈥檙e talking about is virtually limitless wireless bandwidth,鈥� he says. 鈥淰irtually鈥� because so far no one has worked a way of calculating what exactly is the maximum amount of information that can be transmitted over the air waves.
Virtually unlimited bandwidth
There is a theoretical limit, however, described by engineer Claude Shannon in the late 1940s. The Shannon limit of any device is constrained by the bandwidth the device can receive, and how much energy it receives compared with the 鈥渘oise鈥� level in the environment. In practice that limit is huge. 鈥淚t鈥檚 fairly easy to believe that a device that a human can hold (and won鈥檛 get too hot) can cope with something between 100 to 1000 gigabits per second,鈥� says Reed. That sort of signal would broadcast across a wide range of frequencies simultaneously using a technology called Ultra-Wideband (快猫短视频, 24 April, p 28).
But before we reach those heady days of virtually unlimited bandwidth, there are years of fighting to come. It is not just a question of persuading the regulators of the potential benefits of SDR, it is also about convincing licence holders to stomach the change. Reed likens owning a licence to a portion of the spectrum to controlling access to particular shades of colour. 鈥淓ach licensee gets exclusive access to certain colours, and that seems unnatural and unfair to me,鈥� he says. But he faces an uphill struggle to change this.
For starters, many mobile phone operators have paid tens of billions of pounds for exclusive access to portions of the radio spectrum inhabited by 3G mobile phones. And commercial TV and radio companies pay annual fees to broadcast their signals to the public undisturbed. They will not readily give up that control and allow vagabond signals to stroll across their costly turf as software radios search for an empty spot to make a quick call or transmit a signal.
鈥淢ost of the available spectrum has been licensed,鈥� says Paul Kolodzy, a professor of engineering and technology management at the Stevens Institute of Technology in New Jersey. 鈥淏ut most of that licensed spectrum is being used only about 10 per cent of the time.鈥� Reed agrees. 鈥淲hen you look at TV signals, for example, only 25 per cent of the energy in the signal is information, the rest is wasted. Why shouldn鈥檛 someone make sporadic use of that idle space?鈥� he asks.
Such inefficiency also grates on the regulators. Both the FCC and Ofcom are thinking about how to change the status quo. They are preparing a 鈥渟pot鈥� market in spectrum space, where licence holders could sublet, buy or sell portions of their unused spectrum to other companies who want access to the airwaves.
Ofcom is keeping an open mind about SDR. If people could leap in and out of the spectrum without being noticed, says William Webb Ofcom鈥檚 head of research and development, then 鈥渢hat wouldn鈥檛 be harmful, and it should then be possible for regulators to allow software radios even within the current licence structure鈥�.
Instead of a licence specifying which technology and which band it is restricted to, permits would be far more flexible. 鈥淚n the future, that licence might say that you can transmit at this much maximum power in this frequency, your emissions outside of that band can鈥檛 exceed this limit, and your emissions beyond a fixed geographical radius can鈥檛 exceed this level,鈥� says Webb. Calculating those boundaries will depend on how SDR manages to fit in around existing signals in practice.
This means war
In the US, things aren鈥檛 looking quite so easy. 鈥淭he battle lines are being drawn,鈥� says David Rivas, chief technology officer of Sun Microsystems鈥� consumer and mobile systems group. 鈥淢any industries and corporations are concerned that the footing of their businesses will be ripped out from under them.鈥�
Some licensees simply object to having paid billions for exclusive rights only to be told that they may be obliged to share. Others worry that squatters won鈥檛 leave when the licence-holders need their space back. Vetting and balancing such concerns takes time. 鈥淩ules to allow cellular phones began in the late 1960s and weren鈥檛 completed until the 1980s 鈥� and no one was objecting to cellular radio,鈥� says Stagg Newman, a former chief technologist for the FCC.
There鈥檚 an even bigger worry for regulators and licence holders: what happens when the public gets its hands on SDR? 鈥淪uppose someone fiddles with the software and suddenly they鈥檙e operating on the air traffic control frequency?鈥� says Newman. Or it could be something as subtle as tweaking software to broadcast a much stronger signal than regulations allow on a particular frequency, thereby drowning out other transmissions.
Regulators are mulling over three approaches. One is to make sure the hardware devices authenticate the origin of the software; another is to have the receiving device scan downloaded software to ensure it is well-behaved; the third is to design devices so that they鈥檙e physically unable to run software that violates certain rules.
The third approach is Kolodzy鈥檚 favourite. Hackers have so far managed to break through the first two types of defences. 鈥淚t鈥檚 much easier to determine whether something is operating properly,鈥� he says. 鈥淚t鈥檚 like having a governor on your car鈥檚 engine to control speed. The governor can鈥檛 determine if you have the right to drive this car or if you always obey all the rules of the road. It just makes sure you can鈥檛 drive faster than the speed limit.鈥�
That argument leaves some SDR enthusiasts cold. 鈥淚f someone speeds, you give them a ticket; you don鈥檛 limit the capacity of cars,鈥� says Matt Ettus, one of the brains behind an open-source project called GnuRadio. Besides, he says, 鈥渢here already are millions of devices out there that don鈥檛 enforce that. The horse has left the barn鈥�.
GnuRadio鈥檚 mission is to develop software radios and distribute them for free, to show that software radio can鈥檛 be bottled up by corporations or regulatory constraints. Ettus is part of a growing 鈥渙pen spectrum鈥� movement, sparked in part by the advance of software radio technology, that seeks to abolish regulations and open the entire spectrum to anyone who wants to use it. The philosophy: governments license portions of the spectrum to avoid radio 鈥渋nterference鈥�, but if there鈥檚 another way around this bogeyman, why not use it and allow everyone in on the party?
Most observers agree regulators will move towards a structure where frequencies are shared, but the process is going to take a long time 鈥� unless their hand is forced. And Reed says that one little-discussed side effect of GnuRadio could ignite a stick of dynamite under licence holders.
Reed already uses GnuRadio for experiments, and enthusiasts have written decoding software for HDTV, FM and AM radio. There are now efforts being made to create software to decode proprietary formats, including the mobile phones standards such as GSM and UMTS, the technology used in 3G phones. That could open a huge can of worms. Whereas in the past operators had control over who accessed their network because they controlled the production of the hardware, now anyone can emulate their hardware with off-the-shelf electronics. The only protection they really have is through patents on their designs.
This is similar to the problem the software industry faces with pirated copies of its software. 鈥淪topping the distribution of software that infringes your patent is much harder than stopping a manufacturer from making devices that infringe your patent,鈥� says Reed. 鈥淪o people could build their own mobile phones and potentially connect into existing networks without having to buy a device provided by that network.鈥�
Ettus and a team of enthusiasts have designed and built prototypes of a hardware device that can run any GnuRadio-compatible software, called the Universal Software Radio Peripheral. A manufacturer is now in the process of making the first production-line units. Others are sure to follow.
And Ettus isn鈥檛 the first to make such a device. Software radio development company Vanu in Cambridge, Massachusetts, recently demonstrated how existing portable devices are powerful enough to act as software-defined radios. It turned a PDA into a dual-function radio, serving as an FM walkie-talkie and a police-band receiver. And upcoming software will also be able to process mobile phone signals.
It is going to be a long hard fight to free the airwaves, and in the end it will probably be a compromise. Regulation will have to be greatly relaxed to allow these devices to grow. And it is unlikely today鈥檚 licence holders will relinquish control of their turf for a long time yet. But like it or not, software radio is here and it鈥檚 coming to a gadget near you.
