The research and development of baseband is not difficult, but it is very, very difficult. The main reason is that it is difficult to break through the standard essential patent barrier of mobile communication network standards. Baseband is to be applied to mobile phones. As a device accessing the network, mobile phones need to be compatible with other network standards. Taking China as an example, we need to consider 3G GSM, CDMA, 3G WCDMA, EVDO (fortunately, China Mobile has withdrawn from 3G TD-SCDMA), 4TD-LTE and FDD_LTE. When it comes to 5G, we also need to consider NSA and SA access, and all of them are indispensable to be considered as the 5G baseband for qualified mobile phones.
These network standards are formulated by giants in the communication industry, so your baseband can't bypass the necessary patents of these 2/3/4/5G standards anyway. These patents are currently in the hands of communication companies such as Qualcomm, Huawei, Nokia, Ericsson, Samsung and ZTE.
There is a time limit for the standard, but this time is relatively long and it takes 20 years.
In other words, if we do baseband in the 3G era, we have to consider the necessary patents of the 2/3G standard. To do baseband in the 4G era, you have to consider the necessary patents of the 3/4G standard. In the 5G era, you should consider at least 4/5G standard essential patents.
It is a standard essential patent, which makes many chip giants leave, such as Texas Instruments, NVIDIA and Broadcom ... These giants are all stuck in a standard essential patent, and this is just the rise of Qualcomm in the 3G era.
The 3G era is also a problem of CDMA patents. Although Samsung has its own chips, it has to purchase chips from Qualcomm in China, which is the shortcoming of baseband CDMA. It can't be broken. The reason why MediaTek is biased is largely the pot of CDMA.
This is why Huawei began to produce its own chips in the 4/5G era. The biggest reason is that Huawei has mastered a large number of 4/5G standard essential patents, while Qualcomm's 2/3G patents have gradually expired, and they are not so monopolized.
In fact, there is still a very big problem that baseband chips are difficult to make, and it is also difficult for enterprises outside the communication industry to solve. Many mobile communication standards are not very strict standards, and some aspects are just a framework. How to solve this problem depends on the definition of communication manufacturers, that is to say, Huawei, ZTE, Ericsson and Nokia, which produce base station equipment, have different understandings and nuances.
Moreover, there is a very serious problem, that is, each operator's parameter optimization for the network is different, which is even more fatal.
As a result, when you develop a baseband, you must complete the interoperability test with the enterprises that produce base stations in the communication manufacturing industry, and also with the larger operators in the world. This is even more fatal and very difficult for enterprises outside the communication industry.
In a word, baseband production not only faces the necessary patent barriers of communication industry standards, but also equipment manufacturers and operators have a lot of work to do in interoperability testing. It is difficult for non-communication enterprises to do baseband.
Some time ago, Apple publicly stated that it needed 5G baseband, and both Huawei and Qualcomm said that they could sell it. You know, Huawei has always used its own chips and baseband and never sold it, but the relationship between Apple and Qualcomm has long since broken down. Suddenly this phenomenon can be considered that Apple has no choice in the 5G baseband.
Actually, as far as I know, there are only five companies developing 5G baseband, among which China accounts for three. Among them, Huawei and Qualcomm are the best, and the patents in this field are basically monopolized by Huawei and Qualcomm. If Apple doesn't want to lag behind in the 5G signal, it must choose Huawei or Qualcomm. You know, the signal is not very good because Apple gave up Qualcomm baseband and chose Intel baseband because of patent problems. So this time, Huawei and Qualcomm both offered an olive branch like Apple, but from the reality, Apple will definitely choose Qualcomm, which involves many things, and those who are interested can find out for themselves.
The development of 5G baseband is not very slow, but very difficult. Naturally, Qualcomm, as the world's first mobile phone chip manufacturer, can develop 5G baseband, while Huawei started researching 5G years ago. Even so, Huawei has far fewer patents on 5G than Qualcomm, although this is the last choice. As for other manufacturers like Samsung and TSMC, they are only much slower than Huawei and Qualcomm, so it is very difficult to produce 5G baseband.
There has been a lot of news in the communication circle recently. First, Apple was involved in a patent lawsuit with Qualcomm. Intel feels that the progress of R&D is not strong enough to meet the listing plan of Apple's 5G mobile phone. Everyone firmly believes that Apple may not have a suitable 5G baseband chip available. Under this assumption, Apple jumped out to find a 5G baseband chip and knocked on the door until there was no good news.
However, not long ago, the plot began to reverse. Apple not only reached a settlement with Qualcomm and signed a cooperation plan for the next six years, but also the speed of this disagreement is unknown. Then, Intel announced that it would withdraw from the 5G mobile phone modem business and pay more attention to network equipment. Now the latest news, Apple has quoted Intel's mobile phone chip division. If Apple successfully acquires Intel's mobile phone chip division, there will be more news.
After Intel quit, there are only five large-scale commercialization of 5G baseband in the world: Huawei Hisilicon and Ziguang Zhanrui in Chinese mainland, Qualcomm in the United States, Samsung in South Korea and MTK in Taiwan Province Province of China. Why is it so difficult to develop a 5G baseband chip when fewer and fewer manufacturers can afford it in the 5G era?
The research and development of 5G baseband chips is really difficult and the threshold is very high. In the era of 1G network, Motorola is the absolute leader in the communication industry, and its semiconductor technology is also very powerful, which is also the predecessor of Freescale. In the 2G era, many companies have their own baseband chips, such as Ericsson, Nokia, Siemens, Philips and Alcatel in Europe, and TI, Skyworks, ADI, Agere, Broadcom, Marvell and Qualcomm in the United States. Speaking of the 4G era, many companies in Europe and America who can develop baseband chips have withdrawn from this industry, such as Broadcom, TI, Marvell, Nvidia and so on.
1, the chip development time is short. Usually, the development time of a baseband chip is 2-3 years. If we want to compress this cycle, it means more manpower and longer pre-research time. As we all know, last June, the first standard protocol of 5G was frozen. Up to now, there are three mass-produced 5G baseband chips: Qualcomm, Huawei and Samsung, which means that the chips have been commercialized six months after the standard agreement came out. The quality of chips can be guaranteed in such a short time, which requires chip manufacturers to start projects as early as possible, invest money and manpower, and keep up with the pace of 3GPP organizations.
2. Design complexity. The current baseband chips are superior in processing power and lower in power consumption, and the manufacturing process is basically 7 nanometers. The supported network modes increase NR on the basis of the original seven modes, and network diversification also significantly increases the complexity of chip design.
3. The compatibility of RF band is complicated. The RF part of the baseband chip is much more complex, not only supporting 29 new frequency bands of 5G, but also supporting millimeter wave frequency band processing. For millimeter wave, it is necessary to support beamforming and other technologies, which have not been processed before.
To sum up, the research and development of 5G baseband chips should be more difficult than all previous chips, which is why fewer and fewer players can participate in it.
This is a question asked in vain, and it is also the point that many keyboard men don't produce screens or U at home. Let me give you an example, which may not be completely consistent, but it is enough to understand the essential meaning of robbery.
For example, people have to eat, so now Company A has found a method to grow rice and registered a patent. Although it is not difficult to grow rice at the primary level, it is only necessary to find wild species, but others have patents. You can grow your own rice for fun, but you can't sell it. If you want to sell it, you can only buy it from company A, and you can only sell it if you buy the technology. At the same time, Company A developed and registered a series of patents on flour processing, cake making, pie making and bread making. Then all the most easily developed rice-based extension products can only be sold after you pay first. Even if you have the best cooking skills and the best ideas, as long as you sell them, you must pay for each product. How to get rid of this model? The only way not to violate the rules is to plant corn and develop corn products. Then restrict the sale of all corn. Then, plant millet, millet, beans and melons. Now when we find out, there are no new varieties to grow rations. It is almost impossible to find new rations by bypassing known grains such as rice, corn, millet and soybeans. You can only pay, buy seeds and sell them before you pay.
If you want to break this fate, you can only tear your face, kill each other and own the patent rights of these grains.
In the late Qing Dynasty and the Republic of China, internal friction was exhausted and the opportunity to find seeds was missed. Now I can only rub it on the ground. So don't ask why you can't do it in the future, because it's not a question of whether you can do it at all, but a question of whether you can do it. Only by working hard for hundreds of years and generations, waiting for the opportunity and reconstructing the rules of the game! ?
Very difficult!
Think about it, Intel, a company with a market value of more than 200 billion dollars, spent a lot of money to buy baseband business from Infineon. But such a large enterprise will not be able to come up with 5G baseband until at least 2020;
Regardless of how rich Apple is, the baseband can be strangled. Even at the end of 20 18, a department was set up to be responsible for baseband research and development, and even wanted to dig Intel's baseband personnel and prepared for research and development for three years.
Huawei has invested tens of billions of dollars in research and development in 5 G. We must know that Huawei's 4G patents are among the best in the world. Such a huge basic reserve requires such a huge expenditure.
At the same time, in addition to technical difficulties, there are high patent barriers. You want to develop 5G baseband, but you can't abandon all 2G/3G/4G. Each generation of communication system has two standards. If you want to realize full netcom, you need to use a lot of patents. These patents have been held by Qualcomm, Huawei, ZTE, Ericsson, Nokia, Samsung and other enterprises. The use of patents means a fairly high patent fee. Unless it is a patent 20 years ago, the patent for 20 years is quite old.
Therefore, in addition to technology, you also need to authorize the company to pay the patent fee. If you want to bypass these patents, you are wasting your time. The expenses here may need to be calculated according to the cost of hundreds of billions, and the time is in units of years.
At present, there are not many enterprises with 5G in their hands, and even fewer can ship in large quantities. Therefore, 5G R&D needs to face technical problems and patent barriers!
It's hard. NVIDIA, Texas Instruments, stmicroelectronics and Intel. These chip giants have all made baseband chips and failed. Apple also failed to try. Even the chip giant, if there is no strong long-term communication technology accumulation, baseband is dead. Not to mention other companies, it is impossible to throw money hard. Think about it. Apple with the highest cash flow is theoretically the most capable of throwing money, but it can't chew the baseband chip.
Baseband chip is used to process the underlying physical signals, and the most difficult thing is its complex mathematical system. After engineering, the main body is "channel coding and decoding" plus "error correction", which is not something that ordinary chip companies can do. The core design team must have the ability to control mathematics, communication, chip design and implementation at the same time. Therefore, even just looking at design, it is difficult to gather core talents. For example, almost all chip design engineers I have met are directly fragmented at the sight of the word "Fourier". . . .
Some people say that China people can only use metaphors and examples to illustrate. You know, many countries can build planes, but there are fewer third-generation planes. Now there are only three models in two countries, which is getting harder and harder.
It's not difficult at all. Just buy a few patents from Qualcomm and send them back to TSMC to brush a trademark!
There are certainly some difficulties, but it is not a problem for such a large chip research and development enterprise. The problem is that they are commercial enterprises, and profit maximization is king, so whether the cost performance determines whether they will develop, not technical problems! ! !