AI at the IoT Edge is disrupting the industrial market

Transforming data at source aggregation minimizes latency and enables optimal processing of time-critical applications.


Artificial intelligence (AI) at the edge of the network is the cornerstone that will influence the future direction of the technology industry. If artificial intelligence is the engine of change, then semiconductors are the oil driving the new era defined by machine learning (ML), neural networks, 5G connectivity, the emergence of blockchain, digital twins and metaverses.

Despite recent disruptions to the chip industry due to supply chain and, more recently, macroeconomic factors, the confluence of artificial intelligence and the Internet of Things (IoT) known as AIoT is poised to transform the world from cloud-based intelligence to a more distributed intelligence architecture.

It is expected that 73.1 zettabytes of data will be created by IoT devices in 2025, according to IDC research. As a result, endpoint data will increase at a compound annual growth rate of 85% from 2017 to 2025, driving intelligence from the cloud to the endpoint to run AI/ML workloads inside small machines (TinyML). Some of the applications experiencing the most disruption include the development of “voice as a user interface” to improve human-machine communication, as well as environmental sensing, predictive analytics, and maintenance. Key growth sectors include wearables, smart homes, smart cities, and smart industrial automation.

What are the benefits of including intelligence in the endpoint? Many Industrial IoT applications operate in environments constrained by memory capacity, limited computing and battery power, and suboptimal connectivity. Moreover, these applications often require potentially system-critical, real-time responses. Expecting such devices and applications to work in a cloud-centric intelligence architecture just doesn’t work.

This is where the power of embedding intelligence in the endpoint evolves from standard industrial IoT applications to what we call AIoT for industrial applications.

Transforming data at source aggregation minimizes latency and enables optimal processing of time-critical applications. Since the data is not processed and transmitted over the network, security concerns regarding data transmission and flow are greatly reduced. Another advantage is that data processing can be bound to the root of trust at the endpoint, which makes the application impervious to attacks. Because data processing is handled at or near the source, we can take full advantage of the gravity of the data and reduce the energy consumption associated with powering radios or transmitting data over the network.

Our commitment to our customers is to lead the industry in ultimate computing technology with a wide range of MCUs and MPUs. This has already enabled designers to leverage our ecosystem of IoT and AI/ML building by leveraging a technology ecosystem of over 300 commercial grade software blocks provided by trusted Renesas partners.

Our growing AIoT portfolio also explains our recent acquisition of Reality AI, a new AI-powered platform and endpoint in Industrial Internet of Things applications using Renesas processors. Reality AI automatically searches a wide range of signal processing transformations and builds custom machine learning models, while retaining trackability in its approach and delivering valuable hardware design analytics. The modules run on nearly every MCU and MPU core available from Renesas – with new modules being added constantly.

This puts an incredibly powerful tool in the hands of designers to help them solve their toughest problems, because model development is for non-visual sensing use cases and relies on advanced mathematics for signal processing and edge propagation. This enables advanced analytics capable of supporting full hardware design and complete frameworks, including data collection, hardware, firmware, and ML workflows. Other solutions simply generate algorithms and models that often only account for 5% of typical project costs, while ignoring the other 95% of development expenses.

This approach to AIoT design enables developers to reduce unscheduled equipment downtime, improve production efficiencies and perform complex quality assurance tasks that would be costly or difficult to replicate in an existing testing environment.

In a real-world use case tested under 51 different environmental and load conditions in a three-ton residential HVAC system, Reality AI was able to achieve over 95% accuracy when detecting and distinguishing single fault conditions. Testing also detected indoor and outdoor airflow blockages and small charging failures of 5% of OEM specifications in heating and cooling modes.

The convergence of AI and IoT for industrial applications is a huge trend with great potential. Getting Reality AI unlocks the capabilities of combining advanced signal processing and AI at the edge and powered by Renesas hardware, software, tools and ecosystem to provide all the building blocks you need to unleash your creativity.

Sailesh Chittipeddi

(all posts)

Sailesh Chittipeddi is Executive Vice President and General Manager of the IoT and Infrastructure Business Unit at Renesas.

Leave a Comment