The New Full Duplex
Recently, over a period of a handful of weeks there were several announcements regarding full duplex as a next technology. There was this announcement from University of Waterloo, as well as some work by NYU-Poly professor Mike Knox and his team that's part of our collaboration with the university and which was announced and covered by EETimes, among others.
Full duplex is nothing new; most cellular systems today are actually full duplex in that your handset is transmitting and receiving at the same time. But the sticking point is that it is transmitting in one frequency channel and receiving in another frequency channel that's separated quite a bit in frequency. That duplex spacing is what lets your handset transmit and receive at the same time, as well what makes it practical to build a radio that uses relatively inexpensive filters to prevent the transmit signal from leaking into the receiver and completely obscuring the incoming signal.
So, when we talk about the new, full duplex technology, the driver is to be able to transmit and receive on the same frequency. It doesn't take much thinking to realize that this would double the spectral efficiency of the system.
Now, the practical issue is that when your mobile terminal is transmitting, a substantial amount of transmit power is needed to reach the base station, and some small fraction of that power leaks back into the receiver. Although the leakage is a small fraction of the power it's still significantly bigger than the received power from the base station. Various technologies - antenna technologies, RF cancellation technologies, digital cancellation technologies – address that interference, and it's the improvement in these technologies that holds the promise for full duplex today. So, the limitation in full duplex - in fact, let's use "single-channel full duplex" as a term, which is more accurate than just "full duplex" - is really on how much power you can transmit and still maintain a sufficient signal-to-noise ratio relative to the transmit signal that leaks back into your receiver.
The interference mitigation that can lead to single-channel full duplex subdivides into two types; one is temporal or spectral and the other is spatial. Spatial techniques are antenna techniques: multiple antennas, for instance. Some of the architectures use multiple antenna elements with somewhat sophisticated feed networks, but I'd have to say at the moment they're primarily laboratory prototypes. Some of the temporal spectral techniques include RF cancellation and digital echo cancellation-type algorithms.
Now, the problem is usually that the temporal spectral solutions require a measurement receiver that gives you the interference - a sample or a measurement of the interference - which is then either removed in the receive path at RF or at digital baseband. On the cancellation side - both analog and digital technology – the technology is actually out there, and has been used in the past for things like echo cancellation; it just needs to be brought to bear on this full duplex problem and coupled with the antenna technology.
One challenge is that the measurement receiver itself can be a bottleneck, so all of the sorts of typical impairments that exist today in economy-type power optimized receivers and transmitters hinder your ability to accurately measure this really small interference, limiting the ability to completely remove it. With that, improvement becomes an incremental process. In the research literature, people today are publishing 80db of isolation.
How do we get from 80db to the 120db that we need? Largely through trial and error. Most of the basic approaches have been invented; now you have to incrementally improve each of these basic approaches. The engineering challenge – especially the spatial challenge – is obviously amplified by the effort to integrate this into something as small as a handset. Those restrictions don't exist in other applications, which leads us to think that a possible first application of the technology won't be in handsets, but rather in things like cellular relays.
Looking at the eventual benefits, the simple answer is that full duplex in a single frequency channel results in a 2X gain, but of course these things aren't perfect. But there are other benefits. For example, if you think about a cognitive radio space, you're looking for spectrum opportunities. Now, if you really look for spectrum opportunities, it's always much harder to find two bands – a pair – than to just find a single band. The result could be significantly more efficient use of spectrum, beyond the 2X gain in spectral efficiency.
The other benefit beyond spectral efficiency is that the ability to receive and transmit at the same time really means that you can listen and transmit at the same time. This opens the door to new, much more powerful interference cancellation technologies. For example, one of the studies that we've been doing allows not only a single user to transmit and receive at the same time based on the full duplex technology, but also for two different users to use the same spectrum at the same time. The result is new potential ways of delivering connectivity to a group of users.