Activating edge intelligence, how does ADI MCU "show its talents"

I believe that everyone is familiar with MCU products. MCU (Microcontroller Unit) is also a "microcontroller", also known as a micro control unit. In the electronic system, the MCU plays a vital role. It can control other components of the system according to certain procedures, and can also collect some external or internal data for processing, calculation and decision-making. It is well-deserved. The "brain" of an electronic system. In a communication meeting held recently, OFweek Electronic Engineering and many media friends conducted in-depth exchanges with ADI Analog Devices, and further learned about its many innovations and development methods in MCU products.

 

ADI MCU has shipped more than 1 billion pieces so far

 

According to Xin Yi, an application engineer at the China Technical Support Center of ADI, although ADI is famous for its excellent and high-precision signal chain and power supply products, the company has a long history of investing in the field of microcontrollers .

 

Since 1995, the shipment of ADI MCU products has exceeded 1 billion pieces. From the 8051 series of MCUs before 2002, to the MAXQ series of MCUs since 2004, to the ARM core series of MCUs after 2012, ADI's different types of MCUs with different characteristics have left a deep impression on customers. From 2020 to the present, ADI has pioneered and innovated a brand-new edge AI MCU based on the technology of traditional MCUs, so that it can easily realize the artificial intelligence of the Internet of Things in battery-powered devices, which has become a great epoch-making milestone in the ADI microcontroller series. .

 

image.png

 

It is understood that in addition to all the functions of a general-purpose basic microcontroller, ADI's microcontroller products also have excellent performance in many special fields.

 

For example, in terms of power consumption, it has flexible operation modes and ultra-low power consumption, which can greatly improve battery life and extend system life; in terms of interfaces, it has high-speed and safe multi-clock options, which can optimize program performance; in terms of communication, some The product integrates the latest Bluetooth Low Energy (BLE) module, supports features such as long-range mode and BLE audio, allowing devices to communicate over a longer range. At the same time, ADI also provides complete evaluation solutions and development examples for most products. The hardware includes evaluation boards, demonstration kits, etc., and the software includes SDK, library files, driver codes, etc., so as to help users fully understand product performance and accelerate product development. In addition, ADI also has more than 30 years of experience in MCU IP protection and root of trust research and development, which greatly enhances system security and helps security-sensitive equipment manufacturers quickly and efficiently add security encryption, key storage and anti-corruption to their products. Tamper function. In short, ADI's MCU products have six characteristics: low power consumption, excellent interface, new communication, complete evaluation, easy development, and strong security.

 

image.png

 

Xin Yi told OFweek Wike.com Electronic Engineering that ADI's MCU products are mainly divided into three categories according to their functional applications: the first category is low-power MCUs, which have the characteristics of small size, low power consumption, and large storage. The products are relatively general and practical, and are suitable for various products such as industry, Internet of Things, medical care, and consumer products; the second category is security MCU, which has a secure system architecture and strong anti-attack encryption capabilities, which can be used on smart machines or terminals that require high security performance, such as POS machines, card readers, etc.; the third category is artificial intelligence MCU, that is, AI MCU. This type of product is actually born out of the first type of low-power MCU, but because this type of product has artificial intelligence and is not like a general-purpose MCU, it is listed as a separate category. ADI's artificial intelligence MCU is characterized by the ability to push AI reasoning from the cloud to the edge, thereby powering battery-powered artificial intelligence and IoT devices, such as smart homes, face punching, voice control, and more.

 

ADI Edge AI Microcontroller: Helping IoT Practice

 

Back to the theme of this communication meeting "activating edge intelligence". According to Xin Yi, the edge in the "smart edge" refers to the boundary between the physical world and the digital world, such as various temperature sensors, acceleration sensors, acoustic sensors, etc. These sensors are connected to the real physical environment where people live. The world and the invisible and intangible digital world constitute a group of edge nodes. Then the intelligent edge means that these edge nodes also have the characteristics of intelligence, which can provide services such as advanced data analysis, scene perception, real-time decision-making, self-organization and collaboration.

 

At present, thanks to the powerful computing power of AI, AI technology has been widely used in many fields such as industry, medical treatment, and consumption. The IoT technology can use radio frequency, infrared, GPS and other equipment to connect any item to the Internet for information exchange and communication in accordance with the agreed protocol, so as to realize the intelligent identification, positioning, tracking, monitoring and management of the item.

 

However, realizing the powerful computing power of AI will inevitably require a large number of matrix operations, which means a large amount of storage space, powerful computing power, high-speed data interaction, and high cost. Often only large servers deployed in the cloud can do this. Work. Since IoT technology is the interconnection between things, most devices need to be powered by batteries, and the flow of data between things cannot completely rely on the cloud, so the Internet of Things usually means low cost and low power consumption. consumption. This has caused certain contradictions and difficulties in the integration of AI technology and IoT technology.

 

image.png

 

In response to the pain points just mentioned, ADI's edge AI technology combines the advantages of the two to implement AI reasoning at the edge of IoT devices, which is faster than cloud computing. However, edge AI, as AI executed locally on the device, usually needs to be powered by batteries, so it is necessary to strike a better balance between system power consumption, computing speed, and device cost.

 

ADI's edge AI solution MAX7800X series came into being for the new requirements of edge AI. This series of products has a special architecture consisting of two microcontroller cores (ARM Cortex M4F and RISC-V) plus a convolutional neural network (CNN) accelerator. This architecture is highly optimized for the edge, with data loading and initiation being handled by the microcontroller core, while AI inference is exclusively handled by the convolutional neural network accelerator. With the help of the convolutional neural network accelerator integrated in the chip, AI reasoning can be completed without uploading through the Internet. In addition, the microcontroller core is responsible for loading and booting and does not need any operations, so the power consumption is also very low, as long as the battery can be used to power the entire system. With the help of this division of labor between the two hardware, the MAX7800X series does not require networking and supports battery power supply, which perfectly meets the requirements of edge AI and is an ideal product for AIoT applications.

 

In addition, MAX7800X's unique architecture and perfect division of labor, compared with pure software solutions running on low-power microcontrollers, MAX7800X solutions have higher data throughput, speed increased by 100 times, but the cost Just a fraction of an FPGA or GPU solution. Compared with the microcontroller plus DSP solution, the power consumption of the ADI solution is less than one percent of the solution. Therefore, the ADI solution can achieve an optimal balance in terms of power consumption, speed, and cost. Coupled with its small size advantage, it is an ideal product for edge AI applications.

 

It is reported that at present, ADI's edge AI solution MA7800X series mainly includes two artificial intelligence MCUs, MAX78000 and MAX78002, both of which are low-power microprocessors based on Arm Cortex-M4F and RISC-V, equipped with dedicated convolution Neural network accelerators perform AI inferencing. The MAX78000 supports up to 3.5M model weights, while the MAX78002 supports up to 16M model weights. If AI is regarded as a working brain, then the model weight is equivalent to the knowledge stored in the brain. Therefore, compared with MAX78000, MAX78002 can not only realize image recognition, but also perform complex video analysis. In general, ADI's edge AI solution has four characteristics: fast speed, no need for external storage, flexible clock control, and ultra-low power consumption, so it is especially suitable for IoT devices that need to be powered by batteries and need to make timely decisions .

 

ADI Low Power Microcontrollers: Upgrading Consumer & Industrial

 

Xin Yi also mentioned that in recent years, with the high-end development of consumer and industrial applications, users tend to tend to integrate more microcontrollers into the system. These microcontrollers include MCUs for application logic and control, MCUs for sensor data collection, and MCUs for Bluetooth connectivity. At the same time, the system also needs to have an independent power management chip to supply power for these MCUs. However, due to the increasing complexity of applications, smaller and smaller product sizes, and longer and longer battery life requirements, traditional multi-chip solutions are often difficult to meet design requirements.

 

image.png

 

In response to these pain points, ADI's low-power microcontrollers integrate functions traditionally available in multi-chip microcontrollers with a single chip, and also have a built-in power management module. It can be said that ADI's low-power microcontrollers not only feature ultra-low power consumption, but also have the advantages of high performance and abundant resources. Some of them also have built-in Bluetooth modules, so that the size of the system can be further reduced.

 

In terms of industrial applications, ADI has brought MAX32675, MAX32672 and other products, which can be used in industrial flowmeters, optical gas detectors, and MEMS; in consumer applications, ADI’s AMAX32650 is used in wearable smart watches, and can be The MAX32660 used on the true wireless headset, among which, the ultra-small size of the MAX32660 allows it to be used in the compact lidar module of the sweeping robot to analyze the information of the obstacles ahead collected by the sensor. There are also MAX32666 used in wearable ECG T-shirts, MAX32680 used in continuous blood glucose monitoring, and so on.

 

ADI Secure Microcontrollers: Strengthening IP Protection

 

In addition to the Internet of Things, consumption, and industry, security protection is also one of the scenarios that ADI MCU products value. While the electronic payment and Internet of Things markets continue to grow, a large number of devices have also been installed in potentially risky environments, making these Devices are more susceptible to physical attack. These attacks are more threatening than software tampering such as general key cracking and password attacks.

 

image.png

 

In this regard, ADI's secure microcontrollers have achieved strong resistance to physical attacks with keys generated by their dedicated ChipDNA PUF circuits, which can ensure that system-protected keys will not fall into the hands of hackers. The product's flash encryption feature enables reliable IP protection by making sensitive data resistant to the most advanced physical detection methods. Of course, these secure microcontrollers also inherit the characteristics of low power consumption and small size. In short, the secure microcontroller is like an extra protective cover, which locks the data that the user cares about firmly in the protective cover without any leakage.

 

“ADI’s secure microcontrollers can store secure keys, which can prevent various threats through multiple layers of physical security features. For example, we usually use keys and locks to protect our property. If the key is obtained by a thief, then Property is bound to be at risk. Here, ADI's secure microcontroller can add another layer of protection to the 'key'," Xin Yi said, "ADI's products can also resist side-channel attacks. What is a side-channel attack? ?Usually when data is encrypted, the system will produce certain changes, just like when we open the door, there will be a sound. If someone with ulterior motives monitors us all the time, and starts to attack us when we open the door, it will also Incur security risks. ADI's secure microcontrollers can take various measures to disrupt these behaviors, thereby defeating this attack. In addition, our products can also generate high-quality random numbers and have tamper-proof functions, so as to prevent hackers from researching our The laws of encryption, or direct tampering."

 

It is worth mentioning that another major feature of ADI's secure microcontrollers is the hardware solution. Hardware-based solutions can avoid the pitfalls of software solutions, while specialized hardware-based accelerators can run much faster than complex software running on a computer. To sum up, whether it is in the field of electronic cash, smart home, information system or telemedicine, as long as there is a need for data security protection, ADI's products can show their talents.

 

Optimistic about the Chinese market, full of confidence in 2023

 

In a media interview, Li Yong, senior business manager of ADI's MCU product line, said that 2022 may be mainly due to the downturn in some consumer electronics fields, but according to the latest forecast report, the demand for MCUs will increase by 10% from 2020 to 2035. The growth rate has increased, so ADI is still very confident in the future development prospects of the market.

 

As far as the Chinese market is concerned, ADI attaches great importance to the Chinese market, and has a large service support team in China to work closely with local customers. In addition, ADI also has a dedicated Chinese product division, which will research, define, develop, sell and Operate products, and gradually improve the local supply chain and production cooperation system. Li Yong said that ADI is very optimistic about the Chinese market and believes that ADI can work closely with more and more Chinese customers to jointly promote China's industrial digitalization and low-carbon development.