Better article here. One of the biggest advantages is that there is no signal processing on site and therefore no need for a hut at the bottom of the tower. The processing is done at data centres and signal sent to tower via fibre optics. Clustering the baseband units makes it easier for maintenance and also makes it easier to do load balancing across a region. When commuters are driving into work, for instance, the baseband cluster can turn its combined energy to handling the signal load coming from towers along the highways and train lines. During the day, processing could handle heavy downtown traffic, while it shifts focus to the suburbs in the evening. Such load-balancing doesn't produce any additional spectrum or data throughput, but it does mean that a carrier can operate fewer baseband processors, saving the carrier cash.
The connections are fast enough to support a standard called CoMP, or Co-ordinated Multipoint. CoMP, which is currently moving through standardization, relies on the fact that, in many locations, a user's wireless gadget is in range of multiple towers (the closer one comes to the edge of each cell, the more towers can typically see the device). This is usually a waste, since multiple towers spend bandwidth contacting the gadget but can't independently deliver different data. CoMP turns it into a bonus by dividing up requested download data and using all cells in the area to deliver a different slice of it at once—akin to the way BitTorrent operates. The phone then combines the data from all the towers in the proper order. This additive approach to using different towers means that a user's total throughput can go up substantially, but it requires centralized baseband to function.
Finally, the new lightRadio baseband bear can do software-defined protocols. Upgrading to LTE? Just upgrade the software on the baseband processor. (Traditional rack-mounted baseband processors required dedicated units for each protocol.) A new baseband chip from Freescale makes it possible, but it gets even cooler when used in conjunction with the new wideband antennas. LightRadio uses a new antenna that, in Alcatel-Lucent's words, collapses three radios into one. The radios are tiny cubes of 2.5 inches square, and each can operate between 1.8GHz and 2.6GHz. They use tiny amps that can be located atop the tower, built into the antenna enclosure, which keeps the amp size down and dramatically cuts down on the power loss.
These radio cubes are stacked in groups of 8 to 10 in order to make an antenna element, and when one cube in the array goes down, the others remain unaffected. (In a traditional system, the whole antenna unit would fail.) The amps cover enough different frequencies that, in many cases, simply changing the software configuration on the baseband unit can control whether each antenna offers a 2G, 3G, or 4G signal.
The antennas also do "beam forming"—fine-grained directional control over the radio signal—in both the horizontal and vertical dimension to better connect with local wireless devices. Alcatel-Lucent claims capacity improvements of 30 percent through the use of vertical beam-forming alone.
The end result of the system: lightRadio cell towers don't need huts, they don't need air conditioners and heaters, big amps, fans, or even local processing gear. Baseband processing moves closer to the data center model and gets cool new capabilities like CoMP and load-balancing. The system's cost savings come from power (Alcatel-Lucent claims a 50 percent reduction), along with lower construction and site rental fees. The total macro capacity of the system should double while cutting operator costs dramatically.