Text / Chen Jing
Source: Voice of the Storm
Lenovo recalled the history of the 5G coding standard voting two years ago. "Associated Lenovo led to Huawei losing to Qualcomm in the voting." This article hopes to restore some of the truth about the incident. The 3GPP does not have a so-called mechanism to decide whether to win or lose. The abnormal behavior in Lenovo's second meeting did not affect the meeting process. It was not the reason why Huawei's plan failed.
The history of Lenovo's 5G coding standard voting two years ago was revisited by the old. "Associated Lenovo led to Huawei losing to Qualcomm in the voting." Under the background of ongoing trade wars and technological warfare between China and the United States, Lenovo received great public pressure. The founder Liu Chuanzhi has come out publicly.
In fact, there was news about 5G encoding at the end of 2016. At that time, it was claimed that Huawei had fought against Western powers. The dominant Polar code had become an international coding standard. Some people who knew the technology still came up with rumors, saying that it was not the case. The Polar code was not invented by Huawei, and it did not defeat opponents as blown. Unexpectedly, in 2018, it was re-examined, but the argument was reversed. It was said that Huawei's 5G code battle failed and it was associative. What is the truth?
This article explains the ins and outs of the 5G coded voting event and restores the truth. What kind of company is Lenovo? The specific attitude is everyone can choose. But before choosing, you should first understand the truth.
one. What are the options for 5G encoding?
5G is the next generation of wireless communication standards and is in hot development. Wireless communication to receive and send information, the information will be lost or error in the transmission process. What to do wrong? Use coding and decoding to solve. Encoding adds some redundant information. If it fails during decoding, it means that the transmission is wrong and you can request retransmission. Encoding and decoding are algorithmic processes that require hardware devices to perform. According to the different characteristics of various applications, there are many encoding and decoding programs.
The 4G code we use now, called Turbo Code, is dominated by Europe (mainly France Telecom). Why is there only one coding scheme? Because the world needs to communicate with each other, communication companies must also cooperate with each other. If the gauge of the railway in the world is the same, the train can be opened everywhere. However, due to historical factors, countries in the world have several different types of systems, and sometimes it is inconvenient to change vehicles before they can be opened. Global countries and companies are very much concerned with negotiating on communication technology standards. It is not a good thing to engage in a set yourself.
Why can't I continue to use this Turbo code now when I go to 5G? In fact, France Telecom is such a claim, 5G also uses Turbo code. However, the technical characteristics of 4G and 5G are not the same, 5G performance is much better, the transmission rate is more than 10 times higher than 4G, and low latency is also required. Turbo codes are characterized by simple coding but complex decoding and high computational power. Then, in the era of 5G, communication base stations need to decode much more data, but also have low latency. If Turbo code is still used, the required computing power is much higher than in the past, which is uneconomical. Obviously, we must choose a solution with a low decoding speed and low latency to reduce hardware requirements.
It has been found that there are two coding schemes that are better choices. One is an LDPC code (low-density parity check code) and the other is a polar code (polarization code). The LDPC code was put forward in Robert Gallager's doctoral dissertation in 1960, and the Polar code was proposed by Professor Erdal Arikan from Bilken University in Turkey in 2008. And Gallager is Arikan's MIT tutor in the United States.
Erdal Arikan and Robert Gallager
In other words, the Polar code supported by Huawei is not invented by Huawei. Both LDPC and Polar codes originate from U.S. technology, and this has nothing to do with independent innovation. Why does Huawei support Polar code and do not support LDPC code?
A coding scheme does not solve all the problems. It also requires various hardware and software implementations. It requires parallel computing to speed up processing efficiency. It involves a lot of details and many patents. It can be imagined that the LDPC proposed in 1960 had developed earlier and took place much earlier than the Chinese communications companies such as Huawei. Huawei also has some LDPC-related patents, but it is not as good as some of the oldest companies in the industry. When Polar code came out in 2008, Huawei was already the leading company in the industry, and it focused on R&D to accumulate more patents. If the industry chooses to use Polar code for 5G encoding, Huawei will be better at patenting than LDPC. There are also factors in which Huawei has made special input to the Polar code, and several tricks have been prepared. Therefore, Huawei pushed the Polar code as much as the European company pushed Turbo.
However, after all, LDPC has developed for many years and is relatively mature. All kinds of software and hardware support are easy to handle. This is equivalent to 5G coding with a guaranteed choice. The new Polar code theoretically has some advantages (or will not be proposed) and is one of the most important innovations in the industry in recent years. However, after all, the application is still to be developed. One doubt about the Polar code is that its decoding method is called the successive cancellation list decoding, which is not suitable for hardware parallel implementation. The theoretical advantage of Polar code, real hardware, and whether the advantages are not easy to say.
All of the above are data encodings. There is also an encoding requirement in wireless communication, called control channel coding. The control channel is used to transmit commands and synchronization data, which the user does not need to know, but it is also necessary. From the data size, the control channel code block length is generally 20-100 bits, extreme scenes can reach 300 bits; the data channel code block length is 40 to 6000 ~ 8000 bits, and the code block is very large, the amount of data than the control channel It is several orders of magnitude higher. However, the control channel is critical to system performance. It does not mean that the amount of data is small is not important.
The code used for the 4G control channel is called TBCC. At the end of 2016, "Huawei's victory" was actually a result of Huawei successfully promoting the Polar code and replacing TBCC in the 5G standard control channel. At that time, many news companies were not even clear about this, and they began to talk about independent innovation.
In addition, because of the controversy over 5G data channel coding, there are two proper nouns: "Long Code" and "Short Code". Control channel coding is not a short code, and both long code and short code refer to the encoding of data channels. And this is not an academic term. It is because of the conference controversy. Everyone feels vaguely that Polar codes seem to work well in smaller block transmissions (as are the code blocks of the control channel), and LDPC is a better choice for larger code block transmissions. Some people have suggested that why not use a LDPC code for a large code block (long code) and a Polar code code for a small code block (short code)? As for what is the difference between large and small blocks, the meeting did not make it clear that it was dead, but said that the line should be between 128 and 1024 bits, and no specific value has been specified. At the end of 2016, many news misunderstood that Huawei's Polar code won the short code. This is wrong.
European companies that support Turbo code actually mean LDPC for long codes and Turbo code for short codes. Most companies that support Polar code mean LDPC for long code and Polar code for short code. Only Huawei itself believes that the long code and the short code use the Polar code.
Each company supports Turbo, LDPC, and Polar codes based on considerations such as its own patents and industrial chains. How does the international community solve this problem? Do you do it all? This is not possible, then it will be indiscriminate. Products cannot be universal. The same type of equipment must support several types of coding. It can be cumbersome and wasteful. All companies agree that we must discuss a code standard and discuss it with everyone. History has shown that this is an effective method.
So, what exactly is the process of this consultation? This is about talking about Lenovo's controversial 3GPP conference.
two. What is the 3GPP Technical Conference?
First of all, this 3GPP is a telecommunications organization between companies. In theory, there is no national presence. Each company says that it does not involve the country when it says things. Because there are some things, the national interest is not consistent with the interests of the company, or it is troublesome to pull in, and it is not good for solving problems. There are further organizations, such as the IETF (International Internet Engineering Task Force), whose members are individuals and more independent. Companies are not allowed to come in.
From the national point of view, there are also organizations called ITU (International Telecommunication Union), which is an organization under the UN. ITU allocates wireless spectrum and satellite space orbits, formulates global telecommunications standards, provides telecommunication assistance to developing countries, and promotes the development of global telecommunications. In May 2000, TD-SCDMA formally became an integral part of the ITU international standard, and it became the top three 3G international standards along with WCDMA and CDMA2000. TD-SCDMA wants to go to the ITU to say things, this should be recognized by the country. In fact, China is using its own huge market to push TD-SCDMA into the standard. In theory, China does not need the support of other countries and it will do its own internal TD-SCDMA.
Why isn't 5G coded to say what the ITU says? Because coding is a technology implementation, it does not involve spectrum allocation or space allocation. The implementation of such technology has to be specifically done by a number of industry companies. In fact, for companies, they are sometimes more willing to talk about such company organizations as 3GPP. China Mobile is not very willing to use TD-SCDMA 3G, there is no way to have a national mission. China Mobile has been very sulking at 3G for 4G TD-LTE. This is not to say that China should not do TD-SCDMA, but that sometimes, corporate interests and national interests are not consistent.
The 3GPP technical conference related to 5G coding has three times, namely RAN1#86, RAN1#86b, and RAN1#87. This kind of technical conference is not opened as regularly as the World Cup Olympic Games, but there is something to be resolved and the relevant companies come. If a meeting does not solve the problem, it will continue.
How does this kind of meeting open? In principle, it is technically prioritized. It does not represent the state. Each company must theoretically use "technical language" to do business. The 3GPP conference organizer will synthesize the technical opinions of various companies and accept various proposals so that the companies can stand on various proposals. Some proposals with too few supporters will of course give up. For example, European companies say that the 4G Turbo code continues to use the 5G proposal. There are too few supporters and they can be abandoned without any discussion. If there is a proposal for each supporter, discuss it fully.
Why does 3GPP have a vote? This kind of voting is actually a technical nature. It is to look at how many companies each technology supports or opposes. For example, 23 companies support the A program and 24 companies support the B program. Everyone knows that there are many supporters of these two programs and they need to continue discussions. There is a big misconception here that there is no so-called "vote to win".
The 3GPP will not announce a voting framework, saying, "This is the final decisive battle. Those who lose the vote must listen to the winner." There is no such rule. Between companies can't say this, 3GPP still advocates negotiating and solving problems. Industry companies often cooperate with each other, and it's common for them to collaborate. It's not like playing PK ballots. If the voting gap is not large, 3GPP is likely to say: It seems that the differences are larger. This is not a decision beforehand. This is somewhat similar to the trade negotiations. The principle is to find a consensus, not a fixed victory.
Why is voting still very important? Because of the large number of supporters and the large market share, it seems reasonable to say something. If a company’s supporter is then only two or three kittens, and insists that his plan is good, it will appear unprofessional. It does not seem to come to a meeting but it is a mess. Supporters are more powerful and people on this side of the proposal are more likely to persist. We say that our views are such that everyone should do things like that and you should change your mind. Finally, as all agree to a certain plan, the technical meeting is completed. Those who support less can say that we must continue discussions and we convince more companies that this is also possible. However, unless you beat yourself hard to waste manpower and material resources to forcibly support a plan, you will eventually have to discuss it together. Delaying the time is not good for the participating companies. In general, a method is also discussed. Because 3GPP does not allow participating companies to fight against each other to decide whether to win or lose. In general, it is to come together for common interests. The technical plan has been set, and everyone has earned money. In the process, companies have their own patents and the interests of the company can be openly put on the table. Even the national interests can be implicitly reflected. However, in the end, it must be negotiated and resolved.
After the introduction of the 3GPP meeting principles, what exactly is the case with the three 5G coding conferences?
three. The 3GPP three times 5G coding conference process
On August 22-26, 2016, the 3GPP's first major 5G coding conference was held in Gothenburg, Sweden and called the RAN1#86 conference. In April, in Busan, in May, a simple meeting had been held in Nanjing.
Although the industry companies formally talked about 5G coding, they were still exchanging opinions and were called pre-discussions. They did not necessarily say that they must produce results. So the Gothenburg meeting only talked about data coding and didn't talk about control channel coding.
At this meeting, many of the Lenovo voting incidents were ignored. According to the content of this discussion, a group of Chinese companies are like standing in the wrong team.
The public text of this meeting can be downloaded from the 3GPP website at the following address:http://www.3gpp.org/ftp/tsg_ran/WG1_RL1/TSGR1_86/Report/Final_Minutes_report_RAN1%2386_v100.zip
One of these sections says:
R1-167999 WF on ChannelCoding Selection QualcommIncorporated, Samsung, Nokia, ASB, ZTE, MediaTek, Intel, Sharp, MTI, Interdigital, Verizon Wireless, KT Corporation, KDDI, IITH, CEWiT, Reliance-jio, Tejas Networks, Beijing Xinwei Telecom Technology , Vivo, Potevio,WILUS, Sony, Xiaomi
Also supported by Oppo.
Revision of R1-166376
· LDPC should be selected for eMBB data channels toprovide performance and implementation advantages at high rate and largeblocklength
What does this mean? In other words, China's ZTE, VIVO, OPPO, Xiaomi and other companies, and Qualcomm, Samsung, Nokia and other companies support a proposal: Data channels should be encoded using LDPC.
At the same time, there is also a proposal to fight Taiwan:
R1-168040 WF on channel coding selection Huawei, HiSilicon, CMCC, CUCC, Deutsche Telekom, Orange, Telecom Italia, Vodafone, China Unicom, Spreadtrum
Not supported by Orange.
· Polar code is a candidate channel coding technique forNR for eMBB, URLLC, mMTC
This proposal was proposed by Huawei and its own Hass, China Mobile, China Unicom, Deutsche Telekom, Telecom Italia, Vodafone, and Spreadtrum: Let Polar be a candidate coding scheme for data channels. “Not supported by Orange” means that France Telecom will push its own Turbo code and will not support the Polar code.
Looking at the two proposals, this camp is very vague. Huawei has western companies on the side, and several Chinese companies have also participated in Qualcomm and Samsung. In fact, there is no problem of standing in line here. This is a company that expresses its opinions on technology and purely technical means. LDPC, as an old coding scheme, has received support from many foreign companies in China. The Polar code, as an emerging technology, has also received support from some companies and believes it should be a candidate. And it's just pre-discussions, not decisions.
R1-168164 WF on turbocode selection LG Electronics, Ericsson,CATT, NEC, Orange, IMT
· LTE turbo code should be supported for NR for at least low throughput including eMBB, mMTC, and URLLC
· It can support flexibilities of information blocksizes and code rates
· Turbo code enhancement can be considered
· FFS: high throughput
In fact, there is a third proposal. Companies such as LG and Ericsson say that Turbo codes must also be considered for 5G encoding.
At that meeting, Lenovo and Motorola it acquired were doing it? Playing soy sauce. Perhaps because there is no opinion, there is no proposal and no statement.
Because the industry companies seem to have disagreements, the meeting gives a conclusion (or suggestion):
·TheeMBB data channel coding scheme will be chosen at RAN1#86bis
o includingagreeing on the observations that led to the decision.
·Companiesare encouraged to:
o continueanalysis and comparison in order to inform the final decision at RAN1#86bis
o provideany remaining details, especially focusing on LDPC (in view of the situation in this meeting)
o provideany remaining details of the flexibility requirements and how they can besatisfied, and corresponding implementation complexity and any impact onperformance
· Notethat consideration of combinations of coding schemes is not precluded.
·Incase of changes to proposals already available, companies are informing toprovide them at least 1 week before the normal submission deadline for RAN1#86bis.
This means that there will be no disagreement this time, and the data coding scheme will be resolved at the next meeting, RAN1#86b. The party suggested that companies in the industry should actively analyze and compare and give more details to help make the final decision. In other words, nothing was decided.
2. Lisbon Conference
October 10-14, 2016, this proposed 86b conference was held in Lisbon, Portugal. This is also the most discussed meeting in public opinion. The meeting is still discussing the coding of data channels.
As discussed above, after discussion, everyone has come up with an idea and feel that it seems that long and short codes can be handled separately. Each of the long code and short code selects a code that allows companies to express their opinions more fully. This time, Lenovo finally got in.
The text of the meeting can also be downloaded:
One proposal that Lenovo (and its Motorola) participated in was:
Wayforward on eMBB data channel coding
Samsung, Qualcomm Incorporated, Nokia, Alcatel-Lucent Shanghai Bell, Verizon Wireless, KT Corporation, KDDI, ETRI, IITH, IITM, CEWiT, Reliance Jio, Tejas Network, Xilinx, Sony, SK Telecom, Intel Corporation, Sharp, MTI, National Instrument, Motorola Mobility, Lenovo, Cohere Technologies, Acorn Technologies, CableLabs, WILUS Inc, NextNav, ASUSTEK, ITL
Revision of R1-1610689
Also acceptable to Ericsson
· Adopt LDPC code for eMBB data channel as single codingscheme
The proposal is that LDPC codes should be the only encoding for data channels, that is, both long codes and short codes. From the list of companies, Lenovo is indeed very dazzling, the rest are all US, EU, Japan and South Korea companies. It can be seen that there are more companies participating than 86 meetings.
Huawei's proposal is this:
WF onchannel codes
Huawei, HiSilicon, Acer, Bell, CATR, China Unicom, China Telecom, CHTTL, Coolpad, Deutsche Telekom, Etisalat, InterDigital, III, ITRI, MediaTek, Nubia Technology, Nuel, OPPO, Potevio, Spreadtrum, TD Tech, Telus, Vivo , Xiaomi, Xinwei, ZTE, ZTEMicroelectronics
Revision of R1-1610668
Also acceptable to CATT
· Polar code is supported as a channel coding scheme forNR eMBB data channel
The proposal said that Polar code should also be one of the data channel solutions. In fact, it means "long-coded LDPC and short-coded Polar". It can be seen that Chinese companies including Taiwan basically support Huawei, as well as foreign companies such as Deutsche Telekom. ZTE, OPPO, VIVO, and Xiaomi, who supported LDPC in the 86th meeting, all turned. Compared to Lenovo, it does mean that there is something to stand in line with.
The parties looked at the differences and tried to figure out things. A questioning survey was organized to let companies answer:
Question: How many channel coding schemes should be specified for the NR eMBB data channel:
o LDPC: Ericsson, Sony, Sharp, Nokia, ASB, Samsung, Intel, QC, VzW, KT, IITH, IITM, Fujitsu, MotM, Lenovo, KDDI
o Polar: HW
o T+L Accelercomm, IMT, LG, NEC, Fujitsu, Orange
o L+P ZTE, Etisalat, Mediatek, Nubia, Xiaomi, Coolpad, Neul, HW devices,OPPO, CATR, TDTech, Spreadtrum, Potevio, ITRI, IDC, DT, NTU
The question is, should the data channel use several coding schemes? Here, Lenovo and a group of companies said that LDPC should be used as the only choice, which is equivalent to repeating the first proposal. Note that Huawei alone said that the LDPC should be replaced completely with the Polar code. Other companies say that they should use LDPC for long codes, but use Turbo for short codes or Polar.
If the parties are still troublesome, they will continue to organize a "negative vote." Mean to say, give a plan, if you do not agree, say it. The "vote", which is Lenovo's biggest criticism of criticism by public opinion.
- The channel coding scheme for eMBB data is LDPC
No: HW, IDC, HiSi, DT, NEC, CMCC, LG, Spreadtrum, Neul, CATR, Xinwei, TDTech, OPPO, Coolpad, Xiaomi, HW Devices, ITRI, Mediatek, Accelercom, Nubia, IMT, Orange, ZTE, ZTE Microelectronics
- The channel coding scheme for eMBB data is LDPC, atleast for blocks larger than X
- Polar coding is supported for eMBB data for blockssmaller than X
No: Sams, NEC, Intel, QC, LG, Nokia, ASB, MotM, Lenovo, KT, Ericsson, CableLabs, ITL, Sequans, Acorn, Asustek, Mitsubishi, KDDI, Wilus, Accelercom, IMT, Orange, Sony, Sharp, Fujitsu, VzW, Docomo
- The channel coding scheme for eMBB data is LDPC, atleast for blocks larger than X
- Turbo coding is supported for eMBB data for blockssmaller than X
No: HW, IDC, HiSi, Sams, Nok, ASB, KT, QC, Asustek, Spreadtrum, Mitusbishi, CATR, Xinwei, TDTech, OPPO, Intel, Coolpad, Neul,Wilus, Xiaomi, ITRI, Mediatek, Nubia, ZTE, ZTE Microelectronics, HW Devices, CableLabs, ITL, DT, VzW, KDDI, Acorn, Docomo
There are three options here to let people express objections. "Long Code LDPC, short code Turbo" program has the most opponents, 33. “Long Code LDPC, Short Code” Polar” also opposed a number of companies, including 27 companies. This is the voting behavior that has caused Lenovo to get into trouble. Lenovo opposes the Polar code supported by Huawei’s camp. “Long-term short codes use LDPC,” and there are 24 opponents, and Huawei’s camp has voted against it. This is the opinion of some public opinion that Huawei's vote 24:27 was lost to the origin of Qualcomm.
Through this investigation, the conference "consistent with differences" has reached a conclusion:
· Thechannel coding scheme for eMBB data is LDPC, at least for information blocksize > X
· FFSuntil RAN1#87 one of Polar, LDPC, Turbo is supported for information block sizeof eMBB data <= X
o Theselection will focus on all categories of observation, including overallimplementation complexity, regardless of the number of coding schemes in theresulting solution (except if other factors are generally roughly equal)
· The value of X is FFS until RAN1#87, 128 <= X <= 1024 bits, taking complexityinto account
The most important conclusion here is that the LDPC is determined for the long code, and several options for the short code are discussed (discussing what the X value of the short code is also discussed). The basis of this conclusion is that, with the exception of Huawei, all companies agree that LDPC should be used as a long-coded coding scheme. This consensus Huawei does not stop, can only agree. And this has nothing to do with Lenovo's team. Even if Lenovo and ZTE Xiaomi are short-coded for Polar, the long code will be determined to be LDPC.
Below we have to go into the most critical logic discussion. In Lenovo's two opposing proposals, what are the strange positions (that is, the public opinion votes) that are quite different from many Chinese companies? If Lenovo changes and supports Huawei, will the conference conclusions change?
The answer is no effect! This is why many public opinions do not understand.
Some people say that if Lenovo changes to Huawei, 3GPP will decide to use Polar code for short code encoding. This reasoning is absolutely wrong. As the 3GPP resolution is, the short code code is "indeterminate state", the next time the meeting is repeated. Why are you uncertain? Because there are some supporting companies on both sides. There is no voting for winning or losing. As long as the companies on both sides do not have the advantage of being crushed (as all companies have agreed to use LDPC for long codes, only Huawei disagrees), the Lisbon meeting will give an "uncertain" conclusion.
Some public opinion said that Huawei and Qualcomm had 24 and 27 votes each, and Huawei lost. Therefore, Lenovo is a sinner (because Lenovo can pull Motorola to switch to Huawei, so 26:25, Huawei can win). This analysis is absolutely wrong!
First of all, there is nothing that Huawei lost to Qualcomm 24:27. Because the "Qualcomm camp" said that "long code and short code use LDPC", the conference party did not agree. Second, there is no "Lenovo to change investment, Huawei can win 26:25" & rdquo; this matter. Lenovo’s change of investment is nothing more than a slight change in the number of companies opposed by both sides. In the eyes of all participants, it does not have much impact, and there is still a great deal of disagreement. It is absolutely impossible for Lenovo to make a change. "Long-term LDPC, short code Polar" will become the conclusion of the meeting. According to this logic, "Long Code Short Code" uses LDPC & rdquo; should win on the spot with 27:24, but the conclusion of the meeting did not say so. The meeting just allowed everyone to express their opinions fully and gave a conclusion of "Next Time".
It is hard to say clearly why Lenovo's meeting was on the opposite side of Huawei. Lenovo said that it feels more about LDPC and has accumulated more technology. Even if it is to be taken seriously, Lenovo wants to do something bad for Huawei. It wants to rely on the Western camp to express its own "mind to the Western market." This can be speculated that it is personal freedom. However, it must be clearly stated that Lenovo’s voting or standing team did not let Huawei lose opportunities at the Lisbon meeting and did not have any adverse impact on Huawei.
If we have a little understanding of the 3GPP conference mechanism, we can see that without a high level of awareness, this association’s behavior has no effect. Saying that Lenovo's vote has had a key effect, it is lifting it. Unless said that, in addition to Lenovo, there are still a bunch of Chinese companies that have betrayed Huawei, and all of them have supported LDPC on long codes. They do not support Polar as long as Huawei. So Lenovo's vote had a little influence (and with a bunch of other Chinese companies as well), leaving Huawei to lose the opportunity to allow Polar to occupy the long yard (actually impossible).
The above logic may be a bit confusing for those who do not like to use their brains. However, this is not a mathematical problem. As long as one is willing to understand affairs, there is no difficulty in winding.
3. Athens Conference
On November 14-18, 2016, the #87 meeting was held in Athens, Greece. This meeting finally decided the main issue of 5G coding.
Since Lenovo voted exactly the same as Huawei in Athens, there was nothing to say about Lenovo. However, the decision logic in the technical aspects of the 3GPP conference owners is still very interesting.
As already mentioned, "Long Code" uses LDPC & rdquo; This matter has been set, Huawei also gave up. What we want to fight is "What is the use of short code?". Since the Turbo code is the most objectionable person, it will not be discussed. Therefore, the focus of the debate is that "short codes use LDPC or Polar". On this issue, Global Communications Corporation has a large-scale team.
The download address for the conference text is:http://www.3gpp.org/ftp/tsg_ran/WG1_RL1/TSGR1_87/Report/Draft_Minutes_report_RAN1%2387_v010.zip
Huawei organized a camp of 55 companies and demanded that "short codes use Polar". Note that Lenovo and Motorola are on Huawei's side.
WF on channel coding
Huawei, HiSilicon, Acer, ADI, Aeroflex, Alibaba, Bell Mobility, Broadcom, CATR, CATT, Coolpad, Coherent Logix, CHTTL, CMCC, China Telecom, China Unicom, Dish Network, ETISALAT, Fiberhome, Hytera,IAESI, III, Infineon , InterDigital, ITRI, Irdeto, Lenovo, Marvell, MediaTek, Motorola Mobility, National TaiwanUniversity, Netas, Neul, Nubia Technology, OOREDOO, OPPO, Potevio, SGSWireless, Skyworks, Sporton, Spreadtrum, SRTC, Starpoint, STMicroelectronics, TD-Tech, Telekom Research & Development Sdn. Bhd, Telus, Toshiba, TurkTelekom, Union Telephone, Vivo, Xiaomi, Xilinx, Xinwei, ZTE, ZTEMicroelectronics
Bureau Veritas withdrew their support.
·Polar is supported as the channel coding scheme for DLand UL eMBB data with information block up to 1024 bits
Objections: Ericsson, Qualcomm, Nokia, ASB, Samsung, LG, ETRI, KT, VzW, Intel, Docomo, IMT, KDDI, NEC
· Polar is supported as the channel coding scheme for DLand UL eMBB data with information block up to 255 bits
Qualcomm organized 31 camps and demanded that “short codes also use LDPC”.
WF on channel coding for eMBB data Samsung, Acorn Technologies, Alcatel-LucentShanghai Bell, Ceragon Networks, Cohere Technologies, Ericsson, ETRI, EuropeanSpace Agency, HCL Technologies limited, IAESI, Intel Corporation, ITL, KDDI, KTCorporation, Mitsubishi Electric, Motorola Solutions , NextNav, NEC, Nokia,Nomor Research, NTT Docomo, Prisma telecom testing, Qualcomm Incorporated,Reliance Jio, Sharp, SK Telecom, Sony, Straight Path Communications, T-MobileUSA, Verizon Wireless, WILUS Inc
· Adopt LDPC code as the single code for eMBB datachannels
· Adopt LDPC code as the single code for eMBB datachannels at least for blocks >=256 bits
· Adopt LDPC code as the single code for eMBB datachannels
· It is not precluded to employ Polar code as anadditional code for small eMBB data blocks if the concerns on IR HARQ areresolved
· Adopt LDPC code as the single code for eMBB datachannels
· Adopt Polar code for a physical layer control channels
Note that the Qualcomm camp's proposal also contained some concessions to the Polar program. Polar code can be used for small block data in special cases and control channel coding.
In addition to the "data channel short code" controversy, a minor disagreement has also occurred on the control channel, but it is easily resolved. 55 companies such as Huawei proposed Polar code for control channel coding, and five companies such as Qualcomm proposed to use TBCC codes. It is easy to decide that the opinions of 55 companies such as Huawei have become a consensus. The 3GPP meeting decided to use Polar code for control channel coding. This is indeed an achievement of Huawei's camp and is also the focus of many news reports at the end of 2016 (although many say it wrong). The two proposals are as follows. It is clear that there are many more companies in the Huawei camp.
WF on Channel Coding Huawei, HiSilicon, Acer, ADI, Aeroflex, Alibaba, Bell Mobility, Broadcom, CATR, CATT, Coolpad, Coherent Logix, CHTTL, CMCC, China Telecom, China Unicom, Dish Network, ETISALAT, Fiberhome, Hytera,IAESI , III, Infineon, InterDigital, ITRI, Irdeto, Lenovo, Marvell, MediaTek, Motorola Mobility, National TaiwanUniversity, Netas, Neul, Nubia Technology, OOREDOO, OPPO, Potevio, SGSWireless, Skyworks, Sporton, Spreadtrum, SRTC, Starpoint, STMicroelectronics, TD-Tech, Telekom Research & Development Sdn Bhd, Telus, Toshiba, TurkTelekom, Union Telephone, Vivo, Xiaomi, Xinwei, ZTE, ZTE Microelectronics
Bureau Veritas and CGC withdrew their support.
· Polar is supported as the channel coding scheme for DLand UL control channels for eMBB (except FFS for very small payloads)
R1-1613577 WF on coding technique for controlchannel for eMBB LG, AT&T, Ericsson, NEC, Qualcomm
· For DCI, tail-biting convolutional code (TBCC) isadopted as a channel coding technique for NR
· For UCI with encoder input size <=K<=100bits, TBCC is adopted as a channel coding technique for NR
· FFS: enhancements to LTE TBCC including generatorpolynomials with larger constraint length, lower native code rate
So, "The short code uses LDPC, still uses Polar” This question finally is how? According to some people’s voting logic, Huawei’s 55 votes against Qualcomm’s 31 votes won? Unfortunately, the logic of the 3GPP conference is not the same.
The number of 3GPP conferences is not the number of companies, and the weight of large companies is much heavier than that of small companies. There are 31 Qualcomm camps, but there are several big players, such as Samsung, Intel, Ericsson, and Verizon, and their global market share is not small.
Finally, the "consensus" drawn by the Athens conference came about according to the recommendations of the Qualcomm camp: Short codes also use LDPC codes. It may be that the proposal of the Polar channel for the control channel proposed by the Qualcomm camp is for the Huawei camp to give up the confrontation. It can be imagined that all parties to the meeting have argued that Huawei’s plan does not pass, and that the Qualcomm camp has stronger strength, and in the end it has advised the Huawei camp to compromise. The Huawei camp looks at many companies, but it is not very firm. Most companies do not have much technical bias. In the end, things need to be resolved and they cannot be moved. Huawei agreed to use LDPC for short codes and use Polar code as a small gain for control channels.
The Huawei camp cannot be said to lose. Because originally passed the standard, the industry company can develop together according to the standard, each shows the supernatural power to occupy the market. Huawei also has accumulation in LDPC, and will continue to work hard in the direction of LDPC. 5G is certainly still a dominant Chinese company, just as US politicians warned.
four. in conclusion
At the 3rd meeting of 3GPP for 5G encoding, Huawei had unprecedentedly organized a large camp and challenged the US, EU, Japan and South Korea company camps. This is a good practice, indicating that Chinese companies have learned to use "technical language" to express corporate interests, national interests, and the company's vertical integration.
Huawei's main promotion of the Polar code, although not invented by itself, is the first time that Chinese companies have participated in the open competition in accordance with the 3GPP-recognized rules. This is a landmark. Because Huawei's proposal requires the approval of other global companies outside of China, it is not a standard such as TD-SCDMA that can be independently implemented in the country. The Polar code that Huawei promotes, though did not realize the goal that occupy the share in the data channel, but also be considered to be successful in part, become the coding scheme of the control channel.
Lenovo had originally played an irrelevant role in several 5G-coded 3GPP conferences, and it did not attract any attention in 2016. In the third meeting, Lenovo completely joined the camp of Chinese companies and there was no exception. In Lenovo's second meeting, the strange team behavior should be explained by Lenovo. This article points out that 3GPP does not have a so-called mechanism for determining the outcome of winning or losing the game. Lenovo's abnormal behavior at the second meeting did not have any impact on the meeting. It was not because of the failure of Huawei's proposal. Judging from the current pattern of international telecommunications companies, it is unlikely that the Huawei plan will be passed, and it is even more unlikely that it will be passed directly in the second meeting. Regardless of which side of the Lenovo site will not influence, it is not difficult to see from the handling principles of the 3GPP conference.
Public opinion may have some misunderstandings due to the lack of understanding of the technical and process issues involved in telecommunication standards. I hope this article can explain clearly the truth of some events.
(Thanks to communication expert Dr. Yang Xuezhi for providing professional advice on the incident, it is very helpful for this article)