pCell wireless technology

pCell wireless technology

(hopefully) game-changing innovation: pCell wireless technology. If early demonstrations of the technology are to be believed, pCell appears to completely dismantle the idea of a “wireless spectrum crunch,” allowing each mobile device to receive full wireless bandwidth from congested base stations, potentially increasing our total wireless bandwidth by 1,000x. Perlman and his new company, Artemis, are now looking for strategic partners to bring pCell tech to market. We could be looking at one of the biggest revolutions in wireless communications — but there still some big questions about the technology’s efficacy and scalability to answer before we get too excited.

As it stands, there is a very finite amount of people/devices that can share a given piece of wireless spectrum. Say, for example, you’re using your Verizon LTE phone. There is probably 10MHz of available spectrum between your phone and the nearest base station. Using various technologies like beamforming, MIMO, and multiplexing techniques, a bunch of devices can share those 10MHz — but at some point, there are simply too many waves traversing the same frequency, causing game-breaking interference (Shannon’s law). To try and ameliorate this congestion, wireless protocols usually use a “time slot” system, where each user waits in line to send or receive a packet of data. As you would expect, latency climbs and throughput craters. You’ve probably experienced such congestion at conferences and large sports events.

This kind of congestion is wireless networking’s biggest weakness. It can be mitigated by using larger swaths of bandwidth, but there just aren’t that many megahertz available — especially at the all-important lower end of the spectrum, where commercial wireless networks reside. As you can see in the graph above, from the FCC, this problem will only get worse as our thirst for mobile data increases. (Read: New nature-inspired antenna improves wireless performance by 6-8x, coming to routers and smartphones soon.)

Enter pCell, which apparently thrives in these high-congestion, high-interference environments. pCell stands for Personal Cell, but there’s probably some link to the creator (Perlman) as well. pCell is based on a technology called Distributed-Input-Distributed-Output (DIDO) which Perlman has been researching for a while, and first mentioned publicly in 2011. The implementation is fairly complex, but I’ll try to simplify it.

In a standard (simplified) wireless network, you have the end point (your laptop/smartphone/tablet), some kind of wireless access point, and a remote server that you’re trying to access (a website, Spotify, whatever). When you visit a website, the packet goes from your laptop, to the router, to the website — and then the data from the website goes back to the router, and then out to your laptop. In a DIDO network, there is an additional server that sits in the cloud, in a data center, in front of the web server, a special DIDO router (called a pWave), and each device has to have a special DIDO receiver. The DIDO server takes the data from the website and generates a special radio signal that is custom-made for your laptop’s DIDO receiver. This signal then arrives at the pWave, which transmits it as normal, and is picked up by your device.

Normal WiFi (top) where everyone tries to share the same spectrum, vs. DIDO (bottom)

The magic occurs, however, when you have multiple wireless users receiving data from the same DIDO server. Basically, the DIDO server creates special signals that can be transmitted at the same time, rather than one at at time (to prevent interference). These signals, rather than interfering with each other, are actually summed together by the receiving devices. So, if there are 10 devices in the same area, they would receive all 10 DIDO signals at the same time, add them together, and somehow end up with just the data meant for them. Perlman says the maths behind this approach are “immensely complex.” (Read this unrelated but similar story: Increasing wireless network speed by 1000%, by replacing packets with algebra.)

DIDO, in theory, gives every wireless user in a given area full access to the available spectrum, rather than having to share it. In the video at the top of the story, you can see that DIDO seems to work rather well in laboratory testing, but the white paper says, “many of design decisions were specifically made to enable DIDO to be built practically and inexpensively, and to scale to any size.” Perlman says that the first public test of pCell technology will be at Columbia University today. In theory, if DIDO works out, we could be talking about one of the largest revolutions in WiFi and cellular connectivity ever. 

In practice, it very much remains to be seen if pCell will actually work in the wild. As you may have already noticed, in a DIDO setup, every packet of wireless data must go through a DIDO server, a DIDO access point, and be received by a device with a DIDO receiver/antenna. At the moment, as you can see in the video at the top of the story, Perlman and Artemis are using a lot of custom hardware, and computers/servers running Linux and special DIDO software. Perlman maintains that the technology will be cheap to deploy, but that’s kind of beyond the point — in a commercial deployment, almost the entire stack, from data centers, to base stations, to wireless routers, to smartphones, would have to be updated/upgraded to work with DIDO. And I bet Artemis will try to be the sole provider of those DIDO servers.

This is a huge ask, considering Perlman hasn’t even revealed the secret mathematical sauce behind DIDO. It would take a long, long time to work something like this into standards such as 802.11 or LTE. It might not even be possible at all. What we need now is a lot more information from Perlman and Artemis, and then a show of confidence from the industry. We should remain very dubious until a big player like Samsung or Qualcomm signs up. Personally, I think we should be more optimistic about new transmission techniques that don’t require a whole new stack — like infinite-capacity vortex beams.