The media is abuzz with the news of the pCell wireless technology – after all, it’s not too often that someone comes out and claims to have a technology that will change the world! For now, too little has been revealed on this technology, which is understandable for a startup. The aura of mystery is necessary to fuel the hype machine. So, what can we deduce from what little has been revealed? And, what are the prospects of such technology? I will outline here a few thoughts to start this conversation and I look forward to your observations and opinion.
Here are some pointers:
1- Centralized baseband processing: the baseband processing for the pCell is centralized in one location. Centralization enables coordination of distributed remote radios that are placed in different locations. This in turns enables joint processing of signal streams which are transmitted and received by the distributed radios. The principles are similar to Cloud RAN technology where the baseband is pooled and virtualized (doubtful that this is the case yet for the pCell). Joint processing enables features like network MIMO and coordinated multipoint which brings second feature below. For this to work properly the connection between the distributed radio and baseband needs to have high capacity because it transports IQ symbols and it requires very low delay (tens of microseconds) and low jitter (tens of nanoseconds).
2- Network MIMO: LTE base stations feature MIMO antenna systems where on the downlink multiple (mostly two today) data streams are transmitted to the user who’s equipped with two antennas to receive these streams. Provided the signal quality is good, and there’s sufficient isolation between these streams (uncorrelated streams), the receiver will be able to decode each stream and extract the data. This provides a 2x capacity increase. But what if the multiple antennas were not in one location, but are rather distributed across different points in the network? This is possible when the baseband is centralized and connected through high-speed fiber connection to remote radios. The result is similar. The fact that with the pCell the number of antennas N (or antenna locations) needs to match the number of users M to increase capacity by that number is indicative of some form for network MIMO.
3- Beamforming: The demonstration of the pCell shows what Perlman calls a ‘bubble’ around the receiver (1 centimeter in size as he claims). This is indicative of some type of beamforming. There are different ways in which MIMO and beamforming can be combined so the implementation of these features is what matters and, of course, not revealed.
The basic description of the pCell technology and particularly what is termed ‘DIDO’ or distributed input distributed output, is fundamentally the use of uncorrelated multipath to create orthogonal wireless communication channels between a set of transmitters and receivers. In such a case, there is a limit on the number of supported users and their respective capacity that is proportional to the number of antennas, which in this case, includes the number of distributed nodes. In short, there are practical limits to capacity which cannot increase ‘forever’ as portrayed in the media.
With this context, is the pCell a breakthrough in technology? I don’t think so. The principles and mathematics are understood. What is new is that someone actually built and tested it. Furthermore, implementing the technology in practice will be highly dependent on the economics of the link between the central processing and remote radios (called fronthaul). Other practical limitations include the cluster size (number of nodes and antennas that can work in unison) which will impact capacity. To support hundreds of users as LTE does today, there will be no escape, for the time being, from sharing time and frequency resources as it is done now. This is not to take away from what has been accomplished – but to put matter into perspective.
So, these are some of my thoughts to kick start this conversation… I look forward to your observations and opinion!
Great analysis separating facts from fiction! They have actually revealed most of their technology through the patents filings. Check out for example US patent 8571086 and other referenced patents. Interestingly, two ideas/algorithms they use (zero-forcing and block diagonalization) were discussed in a lot of publications around 2004, the same year they filed their patent. That being said there are a lot of obstacles between an idea and a product. If they overcame all the problems kudos to them.
Good analysis, and Rupert Baines has also looked at some the hype in news articles here – http://www.quora.com/Wireless-Technology/Is-pCell-from-Artemis-really-the-Holy-Grail-of-wireless-networking
My major concern is that as you note, 3GPP and others have looked, and are looking at a number of the concepts – multi-cell CoMP, beamforming, massive MIMO, distributed uplink, and nobody has claimed the type of 1000x efficiency gains that are being made for the pCell.
So, big hand to them if they have implemented some of the 3GPP technology ahead of time, but an indication of the secret sauce to justify the efficiency gains would add more credibility
I have looked at a few papers on cell radio technology. Glanced. Really, it would take me hours to really understand the concepts used to wrest every last drop out of the spectrum. I don’t think P-Cell is going to be very far out in front of mobile phone standard setters. By the way, I am interested in that I want to be able to grasp most of what is involved in building personal cell systems such as OpenBTS and its LTE relatives.