My experience with broadcast TV in 1960’s Colorado was fraught with “ghost” images that distorted our television screen during episodes of my favorite TV programs. My parents’ explanation of this being an “echo” from the mountains was very confusing. How could light have an echo? It was in my early days at Hewlett-Packard that I learned the physics of multipath interference; and it was much later that I encountered the technology that would take advantage of these physics rather than fight them.
Multiple In, Multiple Out: This act of adding an advantageous term to the Shannon-Hartley theorem to squeeze a few more bits per second from our precious spectrum is enjoying its highest popularity ever. And while it sounds like a new concept the careful reader will note a reference in Dr. Thomas Marzetta’s 2010 seminal paper on Massive MIMO to a fascinating paper dating from 1919(Alexanderson, Ernst F. W., “Trans-Oceanic Radio Communication”: Transactions of the American Institute of Electrical Engineers Volume XXXVIII, Issue: 2, July 1919). Considerations that smell a lot like MIMO appear to date from a time when the founders of radio communications were but recently in their graves.
Most descriptions of Massive MIMO are either opaque with multi-dimensional calculus, or full of simple brightly colored cartoon diagrams of antennas with lightning bolts. Novices like myself struggle with the topic and there is even significant debate amongst the experts.
MIMO is the use of multiple independent transmit and receive chains each connected to its own antenna to take advantage of the different and independent paths that radio waves follow in a reflective environment. Sophisticated baseband systems split and reassemble signals to and from these different paths to create multiple useful radio communications channels out of what used to be just one. This enables any of the following:
- Use of more than one path to decrease the error rate of a single set of data
- Use of more than one path for different sets of data
- Manipulate the inherent nature of multipath interference to either cancel or emphasize the signal at any physical location in the radio channel
#3 is the essence of what is now called “Massive MIMO”. But based on the heated discussions at industry and academic symposia it is clear there is disagreement about “Massive MIMO” in 5G. A few of the more hotly-debated topics:
Is “Massive MIMO” the same as “Beamforming”? No—as above. MIMO can take advantage of beamforming and indeed FD MIMO has two modes that are strictly referred to as “beamforming” modes. But beamforming is done in many non-MIMO applications.
How many antennas does it take to be “Massive”? Dr. Marzetta stipulates that “Massive” means not only “many antennas” (many more base station antennas than users—and more is always better) but also that each is part of an independent transceiver chain. But the economy of scale given technology available in the 5G time-frame probably means something less than 600.
Is FD-MIMO “Massive”? 3GPP’s FD MIMO introduced in Release 13, has a 64-antenna element count. Hence, many deem it as “massive”. 64 antennas is “much greater than” what??--probably not much more than 10 UE’s. Is FD MIMO really about servicing only 10 UE’s in any one cell? Probably not.
Can you do Massive MIMO in FDD systems? At least one implementation of FD-MIMO in the R13 standard is for FDD scenarios. If one accepts that FD-MIMO is “massive”, the answer to this question is “yes”. But due to the lack of scalability, I do recall Dr. Marzetta stating flatly (and I quote): “FDD is a disaster. End of story.”
Will we get Massive MIMO that will improve capacity, energy efficiency, and spectral efficiency for 5G systems? Yes. MWC 2017 was host to impressive Massive MIMO demonstrations. And the promise of using new digital technologies to take full advantage of a rich radio channel continues to drive innovation. I look forward to it just like I look forward to the next related heated discussion--which perhaps will be a result of this very post.