Millimeter-wave radar technology competition

Millimeter-wave radar technology competition: forward or backward?

Intelligent driving technology has become a commanding height in modern industry and technological competition. In terms of intelligence and applications, China's industry and technology community has done a number of work that has been shown. However, the lack of this aspect does not seem to have attracted enough attention.

The implementation of intelligent driving technology relies on sensors: millimeter-wave radar, lidar, and video systems, among which millimeter-wave radar is the most important. It is noted that, for example, Baidu's self-driving test system uses only lidar and not millimeter-wave radar. We can only think that this is just a test. In the future, commercial systems should be adapted to all weather, and only using lidar will be very limited. This situation should be related to the development of domestic millimeter wave radar.

According to relevant reports in the industry, the research and development situation of domestic automotive radars in recent years is very “lively”: there should be dozens of companies engaged in automotive millimeter-wave radars and investment of more than 10 million, covering automotive, electronics, information-related enterprises, and local High-tech zone; research institutes such as universities, information industry, military industry departments, Chinese Academy of Sciences. However, another situation of domestic development is "embarrassing": the main core components and system technology have not yet made a substantial breakthrough. This situation is worse than the communications-related information industry.

From a business perspective, the civilian radar market is currently much smaller than the communications and network markets, and it is difficult to attract high attention from the industry. From a technical perspective, millimeter-wave radar technology involves more high-end technologies and requires more technical foundation and investment. In fact, people know more about digital chip technology, including CPU, GPU, DSP, FPGA, "Godson", "supercomputing", and AI chips, and so on. There are many practitioners of this type of chip design, and the design difficulty is concentrated at the system level, and there are commercial design tools available. As long as you reach the function module and below, you can usually find the standard library to call and complete it in building blocks. The production process of digital chips is also relatively standardized, and the etch line width can be used to express process procedures and design rules. Therefore, there are many foundry agencies and many fabless (no manufacturing) companies.

The millimeter-wave radar involves a large number of analog chips, such as millimeter-wave chips and system-on-a-chip, and their design and processing are far from being able to be regulated. Commercial microwave circuit design tools are currently only simulation tools, not original design tools. Microwave system-on-chip design requires comprehensive knowledge accumulation of radar system, microwave circuits and components, and process implementation methods to carry out integrated design. For example, microwave circuit unit interconnection requires matching and isolation, while digital unit interconnection design is much looser. From the perspective of production technology, the minimum line width required for millimeter-wave radar chips is not harsh. For example, TI's 79GHz radar chip uses 45nm RFCMOS technology. However, the design of analog microwave circuits requires designers to have a high accumulation of expertise and skills. From a production perspective, in order to achieve product parameter consistency, low noise characteristics, and temperature stability, there are stringent requirements for material purity, process accuracy, and consistency, and there is a high challenge to the yield rate under mass production. Therefore, there are far fewer foreign companies that can provide commercialization of microwave chips, such as TriQuint in the United States, UMS in France, but not in mainland China and Taiwan. If the design is to be produced in a foundry, the designer must thoroughly understand the foundry regulations provided by the manufacturer before designing, and the rules of different manufacturers are very different.

Some people estimate that the overall gap between China's digital chip technology and foreign countries is roughly 3 to 5 years. If this estimate is reasonable, the gap between China ’s analog microwave chip technology and that of foreign countries is doubled.

Vehicle-mounted millimeter-wave radar requires the integration of multiple fields of theory and technology: microwave antennas, microwave integrated circuits, microwave systems on a chip, radar systems, intelligent signal processing, intelligent control, microwave materials, microwave circuit manufacturing equipment and processes, etc., currently there is no one in China. A department or company can basically embrace these technologies. Most of the developer leadership does not understand the difficulty and depth of the problem, and the huge profit temptation has caused some entrepreneurs and developers to have the luck to win. The purpose of this blog post is to alert the relevant parties and try to avoid disorderly competition at a low level. The author has no objection to developers using foreign core chips for application development, but the lack of low-level competition in core technologies is not a long-term solution.

The current limited market and high technical difficulty may be the reason why the millimeter-wave radar has not received large industry and government support so far. With AI technology including intelligent driving becoming a hot topic, millimeter-wave radar and millimeter-wave sensors may enter all aspects of social life in the future, including intelligent driving, intelligent robots, production safety, public safety, and so on. This type of high technology is the cradle of scientific and technological innovation. In the foreseeable future, its application scale may not even be comprehensive in the existing market. So, can we still let go of this? Can it only be bought?

The question now is, will we have obstacles that are difficult to break through in the independent development of millimeter-wave radar technology? The author can say responsibly that we will have many difficulties, but there are no obstacles that are difficult to break through.

Several domestic military research institutes and universities have accumulated rich experience in the past two or three decades. Among overseas returnees in recent years, there has been a skilled elder; the design and production of some microwave chips has begun in China. The current situation is that there are obstacles such as the division of interests and the protection of intellectual property rights in the cooperation or integration between development agencies and technology owners. Loose small circles of industry can be seen everywhere, but it is difficult to form substantive cooperation with complementary technologies.

In this case, it should not be advisable to just increase investment. We need to create a mechanism that breaks through the division of interests, can integrate domestic related technical resources, upgrade the achievements of small workshops to high-end and large-scale industrialization, and integrate the individual achievements into a comprehensive and systematic integration. What can do this can only be companies with strong national behavior or funds. The author proposes that the government take the lead in bringing together the "strength of one skill". The first step is to achieve the complete and independent creation of millimeter-wave radar systems for intelligent driving, to achieve the integrity of product design and manufacturing, automotive-grade reliability and low prices.

The overall development model should be market-oriented, with three core recommendations:

(1) Set up development projects, but do not engage in project bidding and replace them with "technical bidding";

(2) Make detailed regulations and implementation rules on the identification and protection of intellectual property rights;

(3) Speed up the training of professional talents.

The so-called technical tender is aimed at soliciting talent. Individuals at home and abroad can apply regardless of nationality and age, provided that they provide technical results that are directly useful, testable, and independently created for the project. Approved persons enter the work as experts, and their benefits are mainly reflected in the intellectual rights and interests held in the final product. Development agencies at home and abroad may appoint individuals to participate in the bidding, but they need to determine in advance a legal agreement for individuals and institutions to share knowledge and rights. Applicants for theories, methods, and schemes are not excluded, but they need to be innovative, practical, and implementable.

The technical application contains a description of the "technical inspection" method and must be implementable. The tenderer may conduct additional inspections to ensure that the technology meets the requirements of the product target.

Technical tendering highlights the authentic technology itself.

If technical tendering can be implemented, it will necessarily involve the identification and protection of intellectual property rights. Adding "identification" is necessary. The tenderer has the responsibility to protect the intellectual rights and interests of each bidder in the bidding. To this end, the traditional review mechanism must be weakened to give the most voiceful technology creators a chance to perform. The "document registration" system guarantees that original contributions are not overwhelmed or even stolen by "reviews", and provides a legal basis for the preemption of original contributions. Only rigorous intellectual property protection measures can attract creators and ensure the healthy development of technology, and reduce disputes caused by unclear intellectual property rights. Of course, these suggestions need to be further refined, rigorous and operational.

Accelerating the training of professional talents is a fundamental plan. This type of professional technology requires in-depth multi-disciplinary knowledge, and requires basic training conditions for training institutions, which is currently not suitable for most universities. A more feasible approach is to support a number of universities with preliminary conditions, and establish a joint training mechanism for enterprises, microwave integration related research institutes, and universities. Encourage young and interested researchers to challenge difficult subjects. Basic textbooks are not difficult to find. For example, graduated graduate students with electrical engineering and microwave foundations have good guidance and experimental conditions. It should be possible to enter the design role after two or three years of study preparation.

People have realized that technological competition is related to people's well-being and national rejuvenation, and we seem to take it lightly.