How will Edge Artificial Intelligence (AI) Chips Take IoT Devices to the Next Level
Page Contents
Edge Artificial Intelligence (AI) Chips Will Help IoT Devices to Operate Faster and Smarter in 2022
In recent years, edge computing has been gaining popularity to provide IoT (Internet of Things) devices and AI (Artificial Intelligence) applications with valuable sensor information in a fast and efficient manner. But, to effectively implement these innovational technologies at scale, integrated circuit manufacturers and researchers must first build new, specialized chips that can support their computationally-heavy demands.
The Chinese startup Reexen Technology was established in 2018 by Dr. Hongjie LIU, an ETH Zurich graduate. It has now risen to prominence as a worldwide leader in the field of edge-AI ASICs (Application-Specific Integrated Circuits) for medical, industrial, and consumer markets.
Although they have a significant impact on our daily lives, it can be challenging to navigate the vast array of terminology used to describe the latest technical trends, like “edge-AI Application-Specific Integrated Circuits” or “embedded DNN functionalities in low-power internet of things (IoT) sensors.” This article includes two primary goals to solve this problem. The first aim is to give a brief overview of the key concepts related to the emerging field of AIoT (“Artificial Intelligence of Things”), which encompasses many of the terms listed above. The second aim is to provide practical examples of how these technologies are implemented in the real world, using Reexen's work.
IoT (the Internet of Things), Edge Computing, and AI (Artificial Intelligence)
IoT has emerged as one of the most promising new paradigms during the last decade. Broadly defined, “it is simply a network of intelligent objects that can automatically organize and share information, resources, and data. They also can make decisions and respond to changes in the environment.” This widely publicized idea promises to bring everything in our world together under a single infrastructure, allowing us to communicate and connect with anyone from anywhere around the globe. This has led to the proliferation and development of numerous “smart devices” for many sectors, including energy, industrial manufacturing, urban planning, healthcare, etc.
Although there are many definitions for what makes an object “smart,” the most important aspects include their ability to gather information about their environment through embedded sensors. This information must be analyzed promptly. However, large datasets can be quickly generated, especially when many sensor devices are connected within an IoT network. This brings up the question of what type of computing is most suitable for the job.
Cloud computing is the most popular option. It outsources the task of managing, processing, and storing data to a network of remote servers located on the Internet rather than a personal computer or a local server. However, while this strategy is suited for specific IoT sectors, it also has several disadvantages, including decreased bandwidth, increased latency, privacy concerns, and the possibility of data loss.
Therefore, Edge computing emerged as a promising option for time-sensitive applications in which data is analyzed and processed by small computing devices that are located near the data source-i.e., the sensors. These “edge devices” can open up a wide range of applications that use AI, which has resulted in the development of a new field known as AIoT (Artificial Intelligence of Things). This might be a game-changer since industry experts and researchers predict that Artificial Intelligence of Things systems will soon be able to not only identify failures and events but also gather necessary information and make correct decisions based on that data — all without the need for human intervention.
Despite significant advances in this area, several IoT sensor devices still use traditional processor chips. These chips are not well-suited for implementing many of today's computationally intensive algorithmic programs that sensors need to run on the edge. For example, these include DNNs (deep neural networks) and cutting-edge machine learning algorithms responsible for many recent artificial intelligence breakthroughs like DeepMind's AlphaGo. As a result, significant efforts are currently being made to build new ASICs, which, as their names suggest, are mainly designed for a specific application or task. This is where companies like Reexen Technology are attempting to build creative solutions for deploying cutting-edge artificial intelligence technology at scale.
Reexen Technology and Neuromorphic Engineering
As explained by Dr. Hongjie Liu, Reexen Company is involved in neuromorphic engineering- sometimes known as neuromorphic computing. This technique aims to imitate the neural operations and structure of the human brain with hardware and software. Reexen's goal is to mimic the brain's functioning, eye, and cochlea in our ears. This is also called “neuromorphic processing and sensing” they are currently developing “mixed sign in- inference sensing or memory computing” solutions.
Furthermore, mixed-signal in-memory computing circuits solve latency and energy consumption issues in A/D (analog-to-digital) conversion and data-intensive DSPs (digital signal processors) in two ways. First, unlike the traditional CPUs (Central Processing Units) or GPUs (Graphics Processing Units), which can only process data in the “computer-readable” field, mixed-signal computing circuits can process sensory signals directly in both the analog and digital domains. Second, by now integrating computational cells into memory cells, processing-in-memory solutions can solve the shortcomings of the “von Neumann” architecture of traditional computers', which expand significant energy and time to transfer data from memory to the central processing unit for computation.
On the other hand, inference sensing means that inputs generated from the physical world are processed and transformed on the side of the sensors rather than on a large computer or in the cloud, which is beneficial for a variety of applications, including earphones, smartwatches, smart IoT gadgets, etc.
In this case, Reexen Company collaborated with the leading micro-electromechanical systems (MEMS) microphone manufacturer to put its innovative audio-processing chip inside the MEMS sensor itself, allowing the microphone to use keyword detection. This is crucial for many speech recognition applications, including “Hey Alexa,” “Okay Google,” or “Hey Siri.” These words enable digital assistants to respond to users' queries. Reexen technology is currently working on a vision-processing chip mainly used in AR/VR glasses or smartphones.
Conclusion
To summarize, Edge-based computing has proven an appealing solution for IoT devices that provide high-quality and actionable sensor information. It can also save time and reduce energy.
However, industry leaders and researchers have been working together to create new chips that can complete more demanding machine learning tasks on devices in real time-either totally or using a hybrid strategy.
Reexen Technology is a Chinese startup that develops “mixed signal in-memory computing” and “inference sensing” solutions. These solutions aim to mimic the neural operation and structure of the human brain. For example, this has led to the creation of an innovative audio-processing chip used to construct MEMS microphones with integrated keyword spot functions. It is also being used to develop a vision processing chip for AR/VR glasses and smartphones
