Nokia Siemens projects a five-fold increase in spectral efficiency beyond 4G

When wireless industry experts point to huge capacity increases in the future, they’re usually talking about adding carrier bandwidth, not increasing the overall efficiency of the network. Future technologies like LTE-Advanced achieve much of their enormous speeds by layering on more spectrum, aggregating carriers into monster 100 MHz-wide configurations that can support 1 Gb/s connections--they build more lanes in the spectral highway rather than make the cars go faster.

The reason for this is because wireless engineers have started bumping against a limit—called Shannon’s Limit after Bell Labs researcher Claude Shannon—that prevents them from packing any more useful data into a hertz of spectrum before any capacity gains are lost to noise (Telephony: Shannon’s Specter).

With long-term evolution and high-speed packet access plus (HSPA+) technologies, the industry almost reached that limit. Most of the future capacity gains we read about come not from building more efficient pipes, but from building fatter pipes through carrier or from laying many more pipes. Small cells, for instance, mean that the same spectrum can be re-used in more places, while multiple input-multiple output (MIMO) smart antenna technologies send parallel but separate transmissions over the same airwaves.

In a new white paper, however, Nokia Siemens Networks explores technologies and techniques that would boost network spectral efficiency between four and five times than that t achievable in the LTE networks being deployed today, not by overcoming the Shannon’s Limit, but by sidestepping it. According to NSN’s engineers, no individual wireless link can exceed Shannon’s barrier, but the spectral efficiency of the overall system can. By building a network in which multiple cells interact with one another and turn the individual cells’ weaknesses into a system-wide strength, vendors will be able to design networks with an overall spectral efficiency greater than the individual links within it.

The white paper states that typical spectral efficiency on today’s wideband-CDMA networks is between 0.5 and 1 bits-per-second-per hertz (bps/hertz). Extrapolated over a 5 MHz downlink carrier, those numbers line up with the roughly 2.5 Mb/s to 5 Mb/s we’re seeing over HSPA+ networks today. LTE using 2x2 MIMO (such as Verizon’s new network) have a spectral efficiency of just over 2 bps/hertz. Over a 10 MHz downlink carrier that winds up being 20 Mb/s, which lines up with some of the peak speeds reported on Verizon’s LTE network.

But the industry has a veritable grab bag of new technologies that can incrementally boost those speeds. First up is 4x4 MIMO, which could increase spectral efficiency to more than 4 bps/hertz by creating four parallel transmission paths. The obvious conclusion would be to continue doubling up on antennas, gaining a near doubling of capacity with each iteration. But MIMO does have its limits. The antennas must be precisely spaced and shoving more than two smart antennas into a device as small as a handset gets tricky.

From there, NSN moves into system-based techniques, many of which are spelled out in the LTE-Advanced standard. Coordinated multipoint (CoMP) would allow a device to receive transmissions from multiple cells simultaneously, allowing several sites to coordinate the delivery of their data payloads. Relay antennas within cells would shorten transmission distances, increasing capacity at the cell edges (CP: It’s Alive! The birth of the organic network). Inter-cell interference cancellation would further improve that cell-edge performance boosting overall spectral efficiency. Finally, network MIMO techniques could turn the network into a myriad of nodes each transmitting to each other and all of the devices in their vicinity, creating an ‘immersed network’ where there is no such thing as individual cells or cell edges—just a vast collaborative network.

That smorgasbord of technologies would allow networks to boast spectral efficiencies greater than 8 bps/hertz, possibly as high a 10 bps/hertz. To put that in perspective, the 10 MHz downlink carrier used by Verizon’s LTE network could support real-world—not theoretical speeds—of 80 Mb/s to 100 Mb/s.

Considering the constraints of physics, a 5X increase in spectral capacity is nothing to scoff at, but as NSN points out in its white paper a lot more is needed to meet the expected mobile broadband demand of the future.

By 2020, there will be 10 times the amount of mobile broadband subscribers and each subscriber will consume 100 times the bandwidth they do today (about 1 GB per day). That means a 1000X increase in capacity is necessary.
Where will that the additional 200X in capacity come from? The industry will build it the old-fashioned way: by finding more spectrum and building more cell sites.