The ongoing convergence of Artificial Intelligence (AI) and the Internet of Things (IoT), in what’s being referred to as the AIoT, or Artificial Intelligence of Things, has a lot to offer developers across market segments. As connectivity becomes universally interoperable, there will be vast amounts of raw data collected in the IoT. A device equipped with the ability to analyze, learn from, and react to data can help make sense of this and can help make the data deliver more valuable experiences. AI can also help IoT systems be more autonomous, by responding to data in real time, without human intervention.
Though still a relatively new area of the IoT, the AIoT is already changing the way we live and work, by optimizing processes, improving cybersecurity, enabling real-time insights and automating tasks. For example, the AIoT is making smart homes truly intelligent, with systems that become familiar with our habits and preferences and anticipate our needs. In manufacturing, AI-enabled machines are enhancing predictive maintenance and reducing downtime, with the ability to monitor wear and tear, foresee issues and schedule maintenance before actual problems occur. In the supply chain, AI-enhanced logistics and inventory control are making it easier to maintain stocks and fulfill orders and in healthcare, the AIoT is helping caregivers diagnose and treat conditions with greater accuracy. In transportation, AI-enabled systems are reducing congestion and improving traffic flow, and in the enterprise, AIoT devices are monitoring network activity to detect and prevent cyberattacks.
The current selection of AI models tailored for use in the IoT is similarly impressive – covering everything from detecting objects and identifying patterns or anomalies to recognizing keywords, processing natural language or understanding visual information – and new capabilities are being introduced at a rapid pace.
As exciting as these options may be, though, developers should do some advance planning before diving in. Adding AI has the potential to increase the value of just about any IoT device, but entering the AIoT only really makes sense – from a time and effort standpoint – if the addition really contributes to the use case.
Also, when evaluating how to add AI, there are a number of questions to be answered. What connectivity protocol is best suited to transmitting the gathered data? What are the security requirements of the use case? Will my system be able to communicate directly with other devices in the setup? (Hint: this is where Matter comes into the picture.)
The answers to these questions depend heavily on the use case and the ecosystem where the device will operate.
For developers working on devices that will operate in and around the home, answering these questions is simple, because Matter , the new smart-home connectivity standard from the Connectivity Standards Alliance (CSA), is an ideal choice for the foundation of AIoT. Now in its fourth release, Matter spans a range of smart-home device types. Matter was specifically designed to unify consumer devices, networks and protocols so that certified products from any manufacturer can interoperate seamlessly over existing network infrastructures and across multiple smart-home platforms.
Matter supports Ethernet, Wi-Fi, and Thread, which are widely used connectivity protocols, and onboards devices to the network using Bluetooth Low Energy. Matter also defines a common language that creates a securely overlay of the physical IP-based networks. Matter authenticates and authorizes network nodes, creates and manages the secure fabric and defines the structure and semantics of messages that flow within the fabric.
Matter may be a standard designed specifically for smart homes, but the same concepts that define Matter – standardized protocols, security, and semantics – can apply to other domains, too. There are plenty of use cases, in commercial buildings, retail, manufacturing, healthcare and agriculture that have similar requirements and can benefit from the kinds of devices and features, including lighting and HVAC control, that Matter supports. As Matter continues to evolve and more device types are supported, Matter will become all the more relevant to a wide range of AIoT use cases in non-residential verticals.
Having said that, Matter as it exists today has some underlying limitations that tend to rule out large-scale, non-residential deployments. For example, as currently defined, Matter only supports local access, using Local Area Networks (LANs) that operate in unlicensed spectrum. Deployments that use Wide Area Networks (WANs) to support mobility are outside of Matter’s present scope. That includes cellular, satellite and low-power WAN (LP-WAN) technologies, such as LoRaWAN.
Matter is also designed for mass-market verticals, where deployments are managed and secured by consumers. Enterprises tend to operate on a different scale, deploying networks that benefit from being managed and monitored by dedicated IT personnel working with network-management tools. These types of IT-specific management tools aren’t yet available with Matter.
Another aspect of mass-market deployment is that Matter uses compliance certification as the basis for device attestation. That is, when devices are installed, Matter checks the unique attestation certificate of each device to confirm that it is certified by the CSA and built by a trusted manufacturer. This is a good solution for mass-produced consumer products, but it doesn’t address the needs of the more complicated enterprise market, where devices tend to use multi-stage supply chains and undergo extensive customization before being deployed.
Understanding the details of where Matter fits in the AIoT is a more nuanced discussion than I have space for here, but I’d like to recommend a new white paper that thoroughly explores this topic.
Bill Curtis , an analyst specializing in Industrial IoT and IoT technology at Moor Insights & Strategy, has been following the evolution of Matter since its earliest days (in fact, Bill and I first met in 2014 when our companies were founding members of Thread Group) and, as a tech evangelist, he provides a unique perspective on what Matter and AIoT mean for developers.
His latest paper, titled “ AIoT: Connecting AI to the Real World ” discusses how connectivity enables the AIoT, and gives a detailed review of the considerations for AIoT deployment. He explains where Matter fits today and how the Matter approach, which relies on an application layer of standardized protocols, security and semantics, can still benefit verticals that are, as of now, beyond the scope of Matter.
Bill also highlights the NXP approach to Matter and the many ways we simplify and streamline development. NXP has been a significant contributor to the Matter efforts and has a portfolio of Matter-certified platforms with all the silicon, software and services needed to build and deploy the full range of Matter devices, from sensors to gateways and hubs. Bill emphasizes the importance of NXP’s Tri-Radio technology, which combines Wi-Fi, Bluetooth, and Thread on a single chip, and serves to streamline wireless hardware and software development. The result is, as Bill sees it, lower costs, better performance and universal IP-based device connectivity that enable product companies to scale rapidly as the AIoT continues to expand.
We recommend taking a look at Bill’s other research papers, which serve to provide a complete introduction to Matter and its importance in smart-home and consumer-electronic (CE) markets:
We also invite you to discover more about how NXP supports Matter, in the AIoT and elsewhere.
Sujata Neidig has over 30 years of experience in the semiconductor industry and is currently Director of Marketing for Wireless Connectivity where she drives product marketing and standards efforts for IoT connectivity. She also represents NXP on the Thread Group and Connectivity Standards Alliance’s Board of Directors as well as serving as Thread Group’s VP of Marketing. She is based in Austin, TX.