Trying to meet the skyrocketing demand fueled by smart phones and other mobile devices, wireless service providers are striving to introduce new infrastructure while bolstering existing networks.

The new networks being rolled out come in two flavors: WiMax and Long Term Evolution (LTE). The two use similar tricks to allow significant bandwidth increases over the data links used today.

One trick is called orthogonal frequency-division multiplexing (OFDM), which allows a base station to split a chunk of radio spectrum into subchannels. The signal strength of the subchannels and the number of channels assigned to different devices can be varied as needed. OFDM allows high data rates, even far from a base station, and it copes well with the type of radio interference that is common in urban areas, where signals reflect off walls to produce confusing echoes. Both LTE and WiMax also support a technique known as multiple input, multiple output (MIMO), which uses several antennas to create a single wireless connection. MIMO can pack data more densely into the available wireless spectrum than a single-antenna system that uses the same amount of power.

WiMax came to market two years ahead of LTE and offers a theoretical maximum download speed of 144 megabits per second, compared with LTE’s 360 (a typical wired residential broadband connection in the United States runs at around 10 megabits per second). LTE may seem to have a crushing speed advantage, but in practice operators are far from reaching the boundaries of either technology. In the United States, Sprint’s WiMax network offers speeds of three to six megabits per second, while Verizon’s LTE network, which is scheduled to launch before the end of 2010, will offer five to 12 megabits per second. Although Sprint and Verizon are branding these services as 4G, they don’t actually meet the performance criteria that the International Telecommunication Union says officially define a 4G service. But newer versions of both WiMax and LTE are in development that could meet those standards, providing downloads at 1,000 megabits per second or faster.

Although WiMax developed faster (the first large-scale deployments occurred in 2008), most carriers in the United States, Europe, and Japan are using LTE for their next-generation networks. A major reason is that carriers believe LTE will be more technically straightforward to integrate with their established infrastructure.

But some providers, such as T-Mobile in the United States, are avoiding both WiMax and LTE, instead upgrading their existing 3G networks to an improved system called HSPA+. T-Mobile says HSPA+ allows peak download speeds of 21 megabits per second–enough to compete with next-generation networks, at least for now.

In fact, all network operators are trying to boost the capacity and resilience of their 3G data networks as demand soars. Over the next four years, as next-generation systems are built and 3G chips become cheap enough to appear in a wide range of consumer devices, those systems will attract more new subscribers than LTE.

The biggest issue in improving any of these systems is wired networking, not wireless. Cell towers and base stations must be connected to network hubs themselves. These “backhaul” links, traditionally employing copper phone wires or specialized microwave connections, have become bottlenecks in recent years, prompting carriers to shift to more expensive optical-fiber connections. Some 95 percent of the backhaul for Verizon’s LTE network will be fiber when it launches this year. But digging up streets and laying fiber in densely populated places can be expensive and slow. That’s held up AT&T’s efforts to improve its 3G service in New York, which is laboring under the load of thousands of iPhone users.