Spontaneous connections
Bluetooth promises impromptu wireless networking. We consider the potential and the reality.
Paul G Schreier, Contributing Editor

Get ready to experience that rush again, only exponentially. Technology now in place appears poised to drive cordless convenience to a higher plane. Getting rid of those clumsy cords is only the start. Your convergence gadgets will seek out and automatically establish communications with other devices within their vicinity. These self-organizing wireless nets might extend only a few yards, but within that radius, there will be magic.

Some applications are obvious: a wireless headset that allows hands-free use of a cell phone, a cable-free link to the Internet, automatic synchronization between PDAs and their master PCs, automatic exchange of electronic business cards.

But let's brainstorm. With wireless links connecting your electronic arsenal, you could send an instant postcard. You'd snap a shot with your digital camera, add a few lines of text using your PDA, and then send the file off to family and friends using your net-connected mobile phone—all without connecting fussy cables or exchanging memory cards. Or, you might enjoy a single phone handset that functions as a cordless phone at home (taking advantage of fixed-line rates), automatically turns into a digital cell phone when you step out of the house, and morphs into a no-charge walkie-talkie when you get in range of another similarly-enabled phone. Why not use your mobile phone to activate a soft-drink machine and have the charge appear on your next phone bill? Or, what if your PDA could automatically download a map of the Museum of Modern Art as you entered the museum's lobby? You could type in "Starry Night" and let the PDA provide step-by-step directions to locate Van Gogh's masterpiece.

Breakout year

These scenarios are closer to reality than you might think, thanks to localized RF technologies, of which Bluetooth is the most prominent. The phone headset mentioned above is poised for commercial shipment, and the phone-enabled soft-drink machine is running as an experiment. Another firm has patented a sort of electronic sentry: posted in a church or movie theater, it would order all gadgets entering its territory to refrain from loud ringing or beeping. The technology is indeed here.

And the market has staggering potential. Cahners In-Stat Group predicts that in 2003 manufacturers will ship more than 200 million Bluetooth-enabled units, and conservatively expects that amount to triple during the following two years. Research from Frost & Sullivan forecasts western European shipments of Bluetooth chips at 150 million in 2006, generating revenues of around $700 million. The rest of the world should multiply that value by a factor of four or five, elaborates Jan ten Sythoff, the study's author.

But whether all this business actually goes to Bluetooth is still a matter of conjecture. A number of wireless technologies are vying for slices of the market (see sidebar, "Wireless flavors"). Some observers see these schemes as complementary, with each establishing its own niche. Others say Bluetooth's technology and backing will make it an instant dominant force. Truth is, nobody knows for sure, but this is the year when the battle will be joined.

Broad support

Bluetooth has been preparing for its market breakthrough for several years. Ericsson claims to have originated the base scheme in 1994, relates Per-Erik Svensson, that firm's Bluetooth marketing manager. Recognizing the limitations of connectors and cables for mobile phones and accessories, the company set up a research team that devised the RF-based alternative using the 2.4-GHz ISM (industrial, scientific, medical) band, which allows international license-free operation. In 1998, Ericsson set up the Bluetooth SIG (Special Interest Group, www.bluetooth.com) whose other founding members include Nokia, Toshiba, Intel, and IBM. Membership today stands at 1500 and counting.

Bluetooth's operating range, which depends on transmitter power, is nominally 10 meters. An option for a more powerful transmitter extends reach to more than 100 meters, but requires that systems monitor signal strength and adjust their power down to the minimum level needed to maintain communication. This reduction serves two purposes, Svensson explains. It cuts down unnecessary interference with other ISM traffic nearby, and minimizes eavesdropping—would-be spies would have to get fairly close in order to intercept data. The high-power option also requires larger components and incurs more battery drain, inflating costs. In general, these factors restrict the high-power option to larger units such as laptops or access points.

Transparent operation is what truly sets Bluetooth apart from existing wireless-networking technologies. In traditional wired networks, users must predefine the topology. By contrast, Bluetooth supports automatic discovery, in which converging devices sense each other's presence. In other words, you set up an ad-hoc network simply by bringing units together in the same area. Eight devices can join together to form a piconet, and several piconets can combine into a scatternet. Bluetooth supports one asynchronous data channel at 721 kbits/sec, or three simultaneous synchronous voice channels at 64 kbits/sec each, or a channel with both asynchronous data and synchronous voice.

Happy hopping

What happens if two or more Bluetooth piconets try to operate in the same area? Since they use the same frequency band, won't they step on each other's toes?

Bluetooth avoids such network-contention problems and congestion because it's based on frequency-hopping spread-spectrum (FHSS) technology. This means that the carrier (the frequency carrying the data) hops to a new frequency within the selected band 1600 times every second, based on a predetermined sequence. It's as if you and a friend decided to carry on a conversation using a preordained list of pay phones around the city. You'd connect via the first pair of phones, say a few words, switch to another pair of phones, say a few more words, and so on.

In addition to making it hard for eavesdroppers to listen in (more on that in a moment), this allows multiple transmitters to share the same frequency band. Each piconet has its own hopping sequence, so the chances that any two will try to use the same frequency at any given moment are low. Simon Ellis, who serves in a marketing capacity for both Intel's Mobile Computing Group and the Bluetooth SIG, reports some interesting test results. Engineers have set up 10 piconets (a total of 80 devices) in the same radio space and maintained throughput rates of 95 percent of the maximum.

Ericsson selected the 2.4-GHz ISM band because it's open to all comers around the world. But that same advantage has made the band popular with developers of products ranging from baby monitors to garage-door openers to cordless phones. FHSS allows Bluetooth products to peacefully coexist with these devices. Even if a Bluetooth device encounters interference on one frequency, the radio skips to another frequency in mere milliseconds. In addition, the Bluetooth spec allows for several error-correction techniques. However, error correction adds overhead (thereby reducing data rates), so its use will present a tradeoff for designers to judge.

Jan ten Sythoff

Some people speculate that congestion in the 2.4-GHz band might force Bluetooth and other services to another band. "We don't anticipate doing so," reports Ellis. "Intel's long-term view is that because of demand for better speed and performance, wireless LANs, not Bluetooth, will move to another band, perhaps in the 5-GHz spectrum." However, Ellis does believe that as the ISM band gets more crowded, the need to reduce output power will rise. It's similar to what cell-phone companies are doing in congested cities—putting up more transmitters but running them at lower power with shorter range.

The price of success

Obviously, Bluetooth has much going for it. The Bluetooth SIG is mushrooming, and broad industry support helps create market momentum. However, such a multitude of cooks can at least create the danger of spoiling the broth. "Manufacturers turn to Bluetooth for different reasons," says Frost & Sullivan's ten Sythoff. "Some want to optimize the standard for higher speed, or longer range, or interoperability with a new class of devices. Some might see a market niche and fill it with new products that aren't part of the spec. It's unlikely the standard will be able to accommodate all these needs and wants." Even within the Infrared Data Association (IrDA), which has far fewer member companies, interoperability problems have popped up.

Ericsson's Svensson counters that the Bluetooth SIG is well prepared to address this issue, through device qualification and the spec's inclusion of usage profiles. These profiles define a selection of messages and procedures and give an unambiguous description of the air interface for specified services and use cases. Among the first usage profiles are cordless telephony, intercom, headset, dial-up networking, fax, LAN, file transfer, and device synchronization. "We don't want to let interoperability issues destroy the perceived acceptance we've gained so far," notes Svensson. "But there's a big road ahead for Bluetooth to take, and interoperability is critical."

On the threshold

The infrastructure behind Bluetooth is in place, and commercial products are imminent. To bring a device to market, says Intel's Ellis, you need three things: a complete spec, which came almost a year ago, radio chipsets, which multiple manufacturers now offer, and a mechanism to verify interoperability through qualification tests, which the Bluetooth SIG is starting to provide.

OEMs can now purchase radio chips or modules that remove the pain of doing RF design. Interestingly, Ericsson has developed components, including a radio ASIC and a baseband ASIC, but doesn't sell these devices. Instead Ericsson's partners, including sister firm Ericsson Microelectronics and Philips/VLSI Technology, make and sell the components. The company expects to announce other partnerships shortly. Meanwhile, other companies have already entered the component-level business, including Motorola, Cambridge Silicon Radio, Lucent Technologies, and Philsar.

A lack of underlying software, especially drivers for embedded OSs, could initially retard Bluetooth development. In the infrared world, for instance, Extended Systems spent several years porting its IrDA drivers to popular PC and embedded OSs. The firm is among the first to offer a Bluetooth driver, in the form of XTDNDAccess Blue SDK, an embedded protocol stack. This first version, though, only supports the Palm OS, and the SDK initially supports radio chips from Silicon Wave Radio and Ericsson.

As for end-user products, a wireless headset from Ericsson will be among the first to reach store shelves. The 0.75-ounce headset provides hands-free operation of an Ericsson cell phone, which can remain stashed in your pocket or briefcase. Other early products, says Joe Edgerton of Motorola's Personal Area Network Group, will include PC Cards and USB dongles that will allow notebooks to wirelessly connect with cell phones and accessories. Motorola hasn't priced these end-user products, but Edgerton expects that a PC Card might initially sell in the range of $150, with prices dropping over time.

On the PC side, Bluetooth will first appear as an option, Intel's Ellis predicts. New technologies don't become a standard part of the PC until the cost drops below a threshold of roughly $10, he says. Intel hopes to reach that point within two years with Ambler, a Bluetooth module now under development. Meanwhile, the first generation of Bluetooth modules for PCs will cost around $30, so PC manufacturers will offer them as options. Users will be able to add Bluetooth via a plug-in peripheral.

Concerning OEM costs, Ericsson's Svensson cites long-term forecasts of about $5 to incorporate Bluetooth into end-user devices. However, Svensson believes the cost could end up lower than that, if the design capitalizes on functions already built into a system. For example, even mobile phones have lots of digital-signal processing power, he points out; it wouldn't be terribly costly to add Bluetooth radio capabilities into an ASIC.

Wireless flavors In addition to Bluetooth, several other techniques for wireless communications are angling for a spot in the market and in consumers' hearts. In most cases, these technologies differ enough that they won't end up competing head to head. The cable-free concept started several years ago with wireless LANs built around IEEE-802.11, which in its latest version supports data rates to 11 Mbits/sec. Adapters for PCs can run $200 or more. So far, wireless LANs have only found favor with enterprise users, although they look ready to take off in the consumer market (see "Loosely coupled"). INSTANT POSTCARD: Using a Bluetooth-enabled digital camera and a cell phone that features both Bluetooth capabilities and a link to the Internet, you could instantly send a snapshot to friends and family (conceptual photo courtesy of Ericsson). Anyone who's ever used a TV remote or exchanged a virtual business card using a Palm handheld device is familiar with the least-costly wireless technology: infrared. The IrDA (Infrared Data Association) spec has emerged as a cable replacement for electronic devices and peripherals. The OEM cost of adding IrDA functions comes in at less than $2. The spec supports transfer rates up to 4 Mbits/sec, with a 16-Mbit/sec rate in development, but range is limited to about 1 meter. Unlike Bluetooth's radio signal, infrared isn't omnidirectional—the user has to point and shoot within a 30-degree cone. This makes infrared transmissions more secure (more difficult to intercept), but the signal can't penetrate walls. Furthermore, devices must remain stationary during transfers. Although the application spaces for IrDA and Bluetooth overlap, each technology has pluses and minuses. So says Steve Johnson, marketing manager with Extended Systems. Fortunately, he adds, the very scenarios where IrDA falls short are the ones in which Bluetooth excels and vice versa. Bluetooth's only real drawback is its lower data rate. Then there's HomeRF, which like Bluetooth is based on radio and is just now starting to become commercially available. It's a low-speed (1.6 Mbits/sec) subset of 802.11. Like 802.11, HomeRF aims for a role in networking PCs and printers within the home. It's important to note that HomeRF assumes a static topology, with a dedicated station or a PC in charge. In contrast, Bluetooth creates ad-hoc cable-replacement setups that aren't optimized for high-speed data transfers like you'd expect on a network. "The factor that decides the winner isn't which technology is best," comments Dave Sheaffer, director of product marketing at Wind River Systems. "Instead, it's a question of fitting the technology to the market and finding the killer app. Bluetooth and wireless Ethernet have done it, HomeRF hasn't, and IrDA is mature." Sheaffer predicts that Bluetooth will succeed because of who's driving it and how they're driving it. He also sees HomeRF as stuck in a hard place between the price of wireless Ethernet and the performance of Bluetooth. HomeRF has a chance if it can convince end users that they don't care about mobility, but the economies of scale make widespread adoption critical, and interest in HomeRF seems to be fading. Not everyone agrees with this prognosis. In January, Be Inc announced a technology and marketing agreement to integrate HomeRF into BeIA, its version of an embedded OS for Internet appliances. Be chose HomeRF over Bluetooth, claiming the latter is not mature enough. Some companies are hedging their bets. Motorola, for instance, acquired a majority interest in short-range connectivity specialist Digianswer. That firm develops cell-phone accessories and has expertise in both Bluetooth and HomeRF. Motorola in this way hopes to be "transport agnostic"—supporting whatever technology works best for a given application.