Today, we are witnessing an explosion of IoT. Every second, 127 devices are connected to the Internet, and by 2027, the number of IoT devices is expected to reach 43 billion.

As this market grows and evolves, the demand for more sophisticated, powerful, efficient and accurate system solutions to improve the quality of life is also increasing.

Among the many critical technologies that are enabling this exciting future for IoT, edge AI is expected to elevate the capabilities of IoT by enabling data analytics, predictive insights, and intelligent decision-making at the edge.

Let's start with the basics first. So what is edge AI?

Developers and users may be familiar with artificial intelligence (AI) and machine learning (ML), but they may not be as familiar with the term edge AI.

Edge artificial intelligence (Edge AI) refers to implementing and deploying AI applications in edge computing environments or devices close to where data is generated, rather than in a central environment such as a cloud computing facility.

More specifically, edge AI collects data from sensors or other sources (like trackers or health monitoring devices), processes that data using AI models directly on the edge device, and uses the output of those models to take actions or send notifications.

By processing data locally, Edge AI can perform inferences much faster and support real-time use cases, reducing latency and network traffic, improving privacy and security, and increasing energy efficiency.

Today, developers are envisioning a wide range of use cases and applications that leverage edge AI: voice/gesture recognition in appliances, smart home systems, wearables, and health monitoring devices; predictive maintenance in factory automation; and security cameras that can detect and act on suspicious activity in real time. These use cases can be used to increase efficiency and reduce costs.

Representative use cases of AI include smart speakers and voice assistants that utilize voice recognition analysis based on a series of complex AI technologies.

These use cases use Automatic Speech Recognition (ASR) to convert sound waves into words, Natural Language Understanding (NLU) to translate those words into real-world meaning, and the smart speaker uses Natural Language Generation (NGL) to respond.

AI can be leveraged in the smart home market to increase the efficiency of edge devices and provide a seamless user experience.

For washing machines, you can adjust the water level and amount of detergent as well as the number of rinses and spin-dry cycles to make washing more efficient; the thermostat can detect the user's preferred temperature, indoor and outdoor temperatures, and the number of people in the room; the oven can prepare food according to the user's preference and increase safety by ensuring that only adults can use it; and the vacuum cleaner can optimize cleaning depending on the type of floor and increase battery efficiency.

All of these use cases leverage complex AI algorithms at the edge.


New trends and new technologies also bring new challenges to solve. Edge AI is no exception. Edge AI must consider the following aspects in particular:

- High performance (and low power):

The rise of IoT requires more sensors and more information to be shared from various sensors, which leads to increased complexity of devices and the need for more powerful computing performance.

A new generation of edge devices will require high-performance cores and neural network processing accelerators to enable ML operations on the devices themselves.

Adding power optimization requirements to the mix makes things even more challenging, as battery-powered devices need to run efficiently to conserve battery power.

We can reduce energy consumption by storing data and running algorithms on edge devices without having to send everything to the cloud.

- Security and Privacy

Edge AI devices can alleviate security and privacy concerns because they perform most of the operations and data processing themselves and transmit less data to the cloud and external locations.

However, this does not mean that all data on edge AI devices is safe from attacks. As security attacks evolve by the day, edge AI devices require robust embedded security at scale to protect data integrity and privacy.

- Lack of expertise and time required

Developing edge AI devices is nearly impossible without expertise in the field.

This could be a lack of hardware knowledge on how to use specific accelerators or processors for AI/ML, or a lack of expertise on how to use software to develop and deploy AI models.

Lack of knowledge or time-consuming issues can prevent developers from taking the best approach or management from making appropriate decisions.

This is where leveraging comprehensive, robust hardware and software, as well as a complete toolchain solution from experts, can help reduce uncertainty and shorten development times when developing next-generation edge AI devices.

Infineon's PSoC™ Edge is a new family of microcontrollers that addresses the challenges of edge AI, featuring high performance, low power, advanced security, and a range of features to help developers accelerate development time.

As mentioned earlier, increasing system complexity requires higher performance microcontrollers to fuse sensors and process complex data at the edge.A troll is in demand.

In addition, low power consumption and high energy efficiency are important requirements for IoT. To meet these requirements, Infineon introduced a multi-domain architecture to PSoC™ Edge.

This allows for high-performance capabilities such as high-performance cores and hardware-accelerated neural network processor units, while also improving energy efficiency by providing ultra-low-power domains for 'always-on' applications that must be on at all times.

When an IoT device is in deep sleep mode and detects an audio event or facial recognition gesture, the system wakes up, performs the necessary tasks, and then goes back to sleep mode.

Therefore, it maximizes energy efficiency and extends battery life without sacrificing performance. Therefore, edge AI not only accelerates digitalization, but also contributes to decarbonization through power optimization.

Another key challenge is keeping data safe and minimizing security threats.

Therefore, solution providers need to incorporate higher levels of security features into their edge AI products to provide consumers with safer devices.

If not properly secured, edge AI devices can become entry points for attackers to breach the network.

In this case, attackers can also obtain these edge AI-enabled devices that are widely used in the market, such as thermostats, smart speakers, and smart locks, analyze their vulnerabilities, and create malware to compromise the technology and networks.

For these reasons, a robust and appropriately sized embedded security architecture, such as that introduced by PSoC TM Edge, is paramount for new edge AI devices.

Finally, Infineon understands the importance of time-to-market for edge AI devices.

With its recent acquisition of Imagimob, Infineon now has the ability to deliver an end-to-end ML platform with high flexibility and ease of use, with a focus on delivering production-grade ML models.

By providing a complete range of hardware, software and tools, including a robust ecosystem of partners, comprehensive documentation, evaluation kits with connectivity and HMI modules, industry-acclaimed ModusToolbox™ software, and Imagimob’s Edge AI Development Platform and Ready Models, PSoC™ Edge enables developers to reduce development time and get products to market faster.

opages/next-generation-mcu/?utm_source=IOT_world_today&utm_medium=referral&utm_campaign=202312_glob_en_css_css.p.psoc+edge+&utm_content=edge+ai+article" target="_blank">Click here to see how Infineon is enabling the next generation of edge AI devices.