Surprise! Polar codes are coming in from the cold

Polar codes appear to have been accepted into a certain role in 5G, driven perhaps more by political forces than immediate differentiation

As the festive season approaches, there is nothing better than a good surprise. If you follow my blog regularly, you’ll have read my position on Polar codes—great technology, but perhaps not quite ready for prime time.

Well, I guess I was wrong, or maybe not? Recently, 3GPP selected Polar codes as the official coding method for the control channel functions in the 5G enhanced mobile broadband use case (one of the three main use cases being developed), and the LDPC method was crowned as the channel code for the data channels in the same use case. Turbo codes are not in the game yet, at least in this round. So, what happened?

A quick recap: Channel coding

Progress in wireless communications over the last two decades, particularly in link layer technology, has been made possible by advancements and breakthroughs in error-control techniques, i.e. channel coding. This is not an exaggeration.

Channel coding is the most complex and most energy-consuming module of the overall baseband processing unit in your mobile phone. It is quite fair to say that the invention of Turbo codes was one of the key enablers of 3G systems. 4G followed with the inclusion of a reinvention of LDPC codes, making possible the wireless multimedia experience we all mostly enjoy today.

Along comes 5G and the body politic

Since the launch of 5G standardization, we have seen a contentious race (a bloody battle would not be an exaggeration) between the LDPC and Turbo camps at every turn, with each one trying to position itself as the main channel code. Other contenders were not in short supply either. This included some seasoned but old, e.g. new classes of Convolutional codes, and some new but inexperienced, e.g. Polar codes, all of which had been considered to be the underdogs at least for 5G.

As in every popular race, the interested parties—from the supporters of each camp to the technology professionals just watching this race—closely followed the progress and took their positions on either side. Well, the race turned into a historical one this October.

+ Also on Network World: Channel coding: Is there anywhere left to go? +

Before I get into more of this, let me remind you that personally I am a big fan of Polar codes and their potential in 5G and beyond. Polar codes are truly the first explicitly proven codes within implementable complexity that can achieve Shannon capacity. We can thank Professor Erdal Arikan, whose accomplishment here is nothing less than the solving of a 60-year-old mind bender that many have spent their entire careers trying to solve. Professor Arikan’s breakthrough in 2009 created a huge interest in academia.

Despite these shockwaves, when it came to practical implementation and candidacy as a potential player in the nearer term 5G technology game, some challenges in the practical implementation of Polar codes were seen by many as a major roadblock.

Roadblocks however sometimes can magically disappear when politics come into play. This would seem to be the case on Oct. 13, 2016, at the 3GPP meeting where Polar codes were selected following a heavy push from mainly Asian technology players lead by Huawei.

It is important to highlight that Polar codes have been selected for the somewhat less challenging control channel application, whereas the more mature LDPC method was selected for what will be a very high-speed data channel. Regardless, Polar codes are in play, and unless we observe a major turnaround or an unprecedented pushback from major players in the upcoming 3GPP meetings, any 5G-certified mobile cellular technology will operate with a Polar code module/chipset inside. Quite amazing considering these codes were invented less than 10 years ago.

So, what is going on?

Channel codes, modulation and multiplexing schema are where geo-politics and technologies trade off in every standardization generation. Historically, the Turbo code has always been tagged as a European technology and is where most European companies placed their bets (including Orange, which owns many Turbo code patents, and most other major European vendors, including Ericsson and Nokia).

Similarly, though a bit more complex, LDPC has been mostly seen as a North American technology, also enjoying substantial push by major U.S.-based companies—including Qualcomm.

Polar codes are a bit of a strange fish in geo-political terms. They were invented in Turkey (a country that is partly in Europe and partly in Asia). However, they certainly seem to have been adopted as an Asian technology, certainly by the aforementioned Huawei. Huawei has long been a champion of Polar codes (do a quick search on keywords ‘Polar codes’ and ‘Huawei’ to see the scale).

Moreover, the same company has claimed practical implementation of Polar codes with low complexity, with field trial tests achieving 27Gbps in a cellular specific environment. So, maybe this decision shows a swing in 5G to Asia, but let’s not forget that it is LDPC on the data channel. In my opinion, this is a classic standardization compromise. No one would want to put two channel coding methods in a chip if they didn’t have to, but sometimes logic is the first casualty of compromise. This is often the case in the world of international standardization.

What is next?

Now that there is a new player in the game that is clearly not an underdog anymore, the upcoming six to eight months and year will only keep this race very interesting. Don’t forget: This is just round one. There a two other use cases in 5G, ultra-reliable-and-low-latency (URLLC) and machine-type-communications (MTC), and there has been no decision on channel coding here yet. URLLC is particularly interesting with respect to Polar codes because it shares many similar requirements to the eMBB use case.

Personally, I think there is still a lot more work to be done on Polar codes to bring them to their full potential, but we will see what the future brings. Regardless of what happens, I will keep you posted here—so please keep reading.

Copyright © 2016 IDG Communications, Inc.

7 inconvenient truths about the hybrid work trend
Shop Tech Products at Amazon