Zhang Ruian of Espressif: Cloud computing + AIoT is the trend, open source hardware will be more influential

The purpose of the Internet of Things is to connect all "things" to the Internet. These "things" become part of a larger embedded ecosystem in the form of large amounts of data. This process requires developers to create a hardware and software environment for the IoT nodes, as well as application code, and then debug connectivity. Development tools can help designers and speed up this process. Founded in 2008, Espressif Systems is committed to the development of software and hardware products in the field of the Internet of Things, focusing on the research and development of highly integrated, low-power Wi-Fi and Bluetooth MCUs to help the implementation of IoT applications.

Espressif Technology founder and CEO Zhang Ruian recently accepted an interview with EEWeb, a well-known electronic engineering community website. Regarding the development trend of the Internet of Things industry in the coming year, Zhang Ruian is optimistic about the combination of cloud computing with artificial intelligence and the Internet of Things, and mass customization, personalization and artificial intelligence are the key.

Zhang Ruian also pointed out that in the next 20 years, open source hardware will be more influential than it is now. The use of low-power and low-resolution neural networks will be a key feature of open source hardware 20 years from now. For example, using AI modules to replace software will greatly reduce the workload of engineers writing code, and a large part of the design will be based on neural networks. In other words, "software will devour hardware in the form of cheap neural networks."

Born for the Internet of Things

Espressif introduced the ESP8266 and ESP32 in 2014 and 2016 respectively. The ESP8266 is a cost-effective, highly integrated Wi-Fi MCU for IoT applications. The ESP32 is a feature-rich Wi-Fi & Bluetooth MCU suitable for a variety of IoT applications.

Zhang Ruian said that ESP8266 is Espressif's first product designed for the Internet of Things. The product was inspired by the concepts of "One-Laptop-per-Child" and "cheap engineering technology" at the beginning of its design, integrating RF components (except the antenna) into the chip, which is a first in the WiFi field.

ESP32 is Espressif's second attempt to achieve this goal. ESP8266 has many limitations, and ESP32 was developed to solve problems such as memory, computing resources, deep sleep power consumption, and peripheral interfaces. During the design process, the team proposed an asymmetric dual-core architecture. Although this is a small MCU with a non-standard design, the team thought it was a good design and decided to implement it. Therefore, the application of ESP32 is a superset of all the functions of ESP8266.

Zhang Ruian pointed out that Espressif's mission is to make ESP32 a versatile and accessible platform for IoT applications. Espressif's focus is not only on hardware, but also on processors, compilers, operating systems, application frameworks, cloud middleware, development tools, and anything else that can improve the developer experience.

Espressif Systems is currently launching ESP32-C3 and ESP32-S3. ESP32-C3 is a secure, stable, low-power, low-cost IoT chip equipped with a RISC-V 32-bit single-core processor and supports 2.4 GHz Wi-Fi and Bluetooth LE 5.0. ESP32-S3 is an MCU chip with integrated 2.4 GHz Wi-Fi and Bluetooth LE 5.0, supporting long range mode (Long Range). ESP32-S3 is equipped with an Xtensa® 32-bit LX7 dual-core processor with a main frequency of up to 240 MHz, built-in 512 KB SRAM (TCM), 44 programmable GPIO pins and rich communication interfaces.

“ESP32-C3 is Espressif’s first chip that is fully based on RISC-V. It has powerful security features and supports 2.4 GHz Wi-Fi and Bluetooth LE 5.0. We have invested a lot of money to make ESP32-C3 more secure than the first generation ESP32,” said Zhang Ruian. “I am also happy to see security researchers try to crack it, and we will also improve it.”

ESP32-C3 includes several security features such as secure boot, flash encryption, digital signature, HMAC module, and hardware accelerators that support cryptographic algorithms to ensure that the device transmits data securely on the local network and in the cloud.

When talking about why Espressif chose to adopt RISC-V MCU, Zhang Ruian believes that the most important reason is that RISC-V adopts a standardized open source approach and is likely to become the most supported and secure platform on the market in the next few years.

Open source hardware and AI capabilities will become increasingly important

Starting with ESP32-S3, Espressif will launch more cost-effective AI chips in the ESP32-S series in 2021. Zhang Ruian emphasized that the algorithm of AI chips is as important as the hardware.

He pointed out that it is difficult for ordinary users to collect enough samples and train their own algorithm models at this stage. Therefore, Espressif will provide rich AI codes on GitHub for users to use directly. At the same time, the company has conducted a lot of tests internally and has now achieved excellent beamforming, wake-up word recognition, and offline command word recognition functions on ESP32-S3.

In the near future, ESP32-S3 will also have powerful image recognition and processing capabilities, and the AI ​​computing capabilities of ESP32-S4 will be further upgraded.

In the foreseeable year of 2021, Zhang Ruian believes that cloud computing will be combined with artificial intelligence and the Internet of Things, with mass customization, personalization and artificial intelligence being the key ingredients.

In the unforeseeable future, Zhang Ruian believes that open source hardware will be more influential in the next 20 years than it is now. Zhang Ruian believes there are three reasons for this. First, technological advances will make rapid prototyping (RP) possible at a fraction of the cost of today. Second, global environmental issues will bring more troubles that may involve only a few or no large number of business priorities, so open source hardware is particularly important. Third, with the rapid development of artificial intelligence hardware and algorithms, and the availability of computing resources, manufacturers are able to create almost anything with very simple hardware. For example, manufacturers can make their own communication baseband, consisting of a simple radio and a neural network-based modem, which is much more power efficient than existing ones.

Zhang Ruian also pointed out that a key feature of open source hardware 20 years from now is the use of low-power and low-resolution neural networks. By 2040, in most cases, time-consuming work such as writing code will be greatly reduced by replacing software with AI modules. In short, software will "swallow" hardware in the form of cheap neural networks.