Alas, the enthusiasm for 4G isn't due to accelerated progress; it's because third-generation (3G) services have proven so disappointing. Instead of one standard worldwide, there are three incompatible systems in the United States alone. Voice is carried over the circuit-switched infrastructure inherited from second-generation (2G), not the promised IP. The touted streaming video is just a low-resolution slideshow. Most importantly, the data rates are closer to dial up than DSL.

This is partially due to the technology's immaturity: 3G systems rolled out so far could be considered beta versions, with the real thing still in the future. But 3G will never live up to its creators' promises. Despite early excitement about data, the main economic incentive for 3G is increased capacity for narrowband voice. Though data rates will increase, there isn't enough bandwidth to transfer large e-mail attachments quickly, let alone stream audio or video at broadcast quality as the cell phone vendors first claimed.

If you believe the industry, 4G will enable all this and more: Many companies talk of holophones, remote-controlled cars, and mobile virtual reality. Given their past record of hype, there are good reasons not to believe the more outlandish predictions, but there are also reasons to think that some aspects of 4G could be real.

2020 VISION
Despite its apparent novelty, 4G has been in gestation for more than a decade. The first research took place in Europe during the early 1990s, and was intended to investigate very high data rate technologies that could serve the needs of mobile communications until 2020.

The most advanced project was the Mobile Broadband System (MBS), a collaboration between several companies and universities overseen by the European Commission. MBS' inventors intended to create a cellular system with low latency, guaranteed QoS, and a data rate of OC-3 (155Mbits/sec)-thousands of times faster than anything available today, let alone ten years ago. Incredibly, they succeeded.

Built in 1995, the MBS prototype had a data rate of "only" E3 (34Mbits/sec), though higher rates could be achieved by using several links in parallel. It was tested in several indoor and outdoor environments, including driving around a city block at around 30 miles per hour. MBS' physical layer was based on a variant of Time Division Multiple Access (TDMA), used by most 2G phones, and its higher layers were based on ATM-at that time, still considered the networking protocol of the future.

MBS' inventors recognized that a prototype isn't the same as a functional network, estimating that the system would take 15 years to commercialize: The first deployments would be in 2010, with widespread services by 2020. Halfway through that time, the specification has changed. The physical layer will be based on Orthogonal Frequency Division Multiplexing (OFDM), a technology designed to resist interference caused by a radio signal's multiple paths. And ATM has been abandoned in favor of IPv6.

While other wireless generations have focused on voice, perhaps combined with other types of traffic such as short messages, MBS is intended to be service-independent: It's a big pipe for data, which devices can use for any applications. Japan is so far the only country where mobile data has proven popular and profitable, so it's no surprise that Japanese operators are leading the race to 4G.

Instead, i-mode made DoCoMo the world's largest ISP. The company hopes that true mobile broadband will enable it to replace fixed access entirely, and it plans to have a 4G system operational by 2006. Ono also predicts that the future of mobile data is more than Web surfing and access to corporate servers. DoCoMo thinks there's a big market for "machine-to-machine" communication: vehicles that take directions from a central traffic computer, or domestic appliances that automatically reorder supplies.

DoCoMo also hopes to improve 3G systems so they reach the hyped data rates. Vendors once claimed that 3G would enable full duplex access at 2Mbits/sec, but every real system built so far has an absolute maximum speed of 384Kbits/sec downstream, 64Kbits/sec upstream. "In the near future [2004], we will have 2Mbit/sec access," says Ono.

Just as current upgrades that add packet switching to digital cellular systems are known as "2.5G," improved versions of 3G are often called "3.5G." The upgrade closest to reality, set to be standardized in 2002, is called High Speed Downlink Packet Access (HSDPA). This uses better modulation techniques to reach up to 10Mbits/sec. All users in a cell share the capacity, but in a very efficient way called "extreme unfairness," which gives more bandwidth to people in interference-free areas instead of trying to portion it out equally.

HSDPA only works with Wideband Code Division Multiple Access (W-CDMA), the worldwide 3G standard developed in Europe. Many American operators are instead using narrower-band CDMA systems, which are theoretically more spectrally efficient. (See "Waiting for Wireless in the United States," December 2001, page 54.) They achieve this efficiency by using the same modulation techniques as HSDPA, leaving no room for expansion.