AI accelerator boosts Matter-ready IoT chip

AI accelerator boosts Matter-ready IoT chip
Technology News |
Multiprotocol 2.4 GHz Gecko wireless SoCs from Silicon Labs improve AI/ML Tensorflow performance via accelerator block and 20bit ADC
By Nick Flaherty


Silicon Labs has added an accelerator to its multiprotocol 2.4GHz wireless chips to add machine learning capabilities to battery-power designs.

The accelerator in the BG24 and MG24 Gecko chips is as yet unspecified but natively supports TensorFlow models, says Ross Sabolcik, VP and general manager of Industrial and Commercial IoT at Silicon Labs.

This is important for applications such as ‘wake word’ recognition as well as signal processing such as recognizing sounds through a new software development kit. Signal analysis code such as TinyML can still run on the 78MHz ARM Cortex-M33 processor, he says.

“This is really the culmination of the work we have been doing for the last decade with our most capable radio with a 20dBm output and current consumption sub 4mA for receive,” he said. “One of the big things that is driving the memory is Matter and the field upgrade and you need a big flash for OTA.”

The flash and SRAM in the multiprotocol chip has been enlarged to 1536kB and 256kB respectively to support over the air (OTA) updates. The chips also include a 20bit analog-to-digital converter (ADC) for more accurate sounds and signal analysis.

For example, in a commercial office building, many lights are controlled by motion detectors that monitor occupancy to determine if the lights should be on or off. However, when typing at a desk with motion limited to hands and fingers, workers may be left in the dark when motion sensors alone cannot recognize their presence. By connecting audio sensors with motion detectors through the Matter application layer, the additional audio data, such as the sound of typing, can be run through a machine-learning algorithm to allow the lighting system to make a more informed decision about whether the lights should be on or off.

The co-optimized hardware and software platform will help bring AI/ML applications and wireless high performance to battery-powered edge devices. The chips support multiple wireless protocols through an ARM Cortex-M0+ core, including Matter and Bluetooth, and incorporates PSA Level 3 Secure Vault protection for smart home, medical and industrial applications.

“We have added an AI/ML separate accelerator and when we benchmark TensorFlowlite on the accelerator we see a 4x reduction in power and a 6x improvement in speed,” said Sabolcik. “We’ve always made wireless SOCs with a comprehensive software stack and you can take the AI models and push them directly onto the SoC but it’s still early days in getting to the point of offering AI stacks. The other important peripheral is security at PSA level 3 with the secure vault and you can see that we will have to do more and more on the security side. For example we have Custom Programmable Manufacturing service, where you can take an image [of software] that you have signed and give us a key and we program the image into the device and lock the bits to turn on all the security.”

Next: TinyML tool partnerships

In addition to natively supporting TensorFlow, Silicon Labs has partnered with some of the leading AI and ML tools providers, like SensiML and Edge Impulse, to ensure that developers have an end-to-end toolchain that simplifies the development of machine learning models optimized for embedded deployments of wireless applications.

More than 40 companies representing various industries and applications have already begun developing and testing this new platform solution in a closed Alpha program says Sabolcik.

For example, global retailers are looking to improve the in-store shopping experience with more accurate asset tracking, real-time price updating, and other uses. Participants from the commercial building management sector are exploring how to make their building systems, including lighting and HVAC, more intelligent to lower owners’ costs and reduce their environmental footprint. Finally, consumer and smart home solution providers are working to make it easier to connect various devices and expand the way they interact to bring innovative new features and services to consumers.

The EFR32BG24 and EFR32MG24 SoCs in 5 mm x 5 mm QFN40 and 6 mm x 6 mm QFN48 packages are shipping today to Alpha customers and will be available for mass deployment in April 2022. Multiple evaluation boards are available to designers developing applications. Modules based on the BG24 and MG24 SoCs will be available in the second half of 2022.;

Related articles

Other articles on eeNews Europe

Linked Articles
eeNews Europe