Technology company Silicon Labs has recently announced the MG24 and BG24 families of wireless SoCs designed for Multiple-protocol and Bluetooth operations respectively, and a new software toolkit. This co-optimized software and hardware platform will bring AI/ML capabilities and wireless high performance to edge devices powered by the battery.
The 2.4-GHz MG24 and BG24 families will support wireless protocols and incorporate the PSA Level 3 Secure Vault™ protection, which is ideal for medical, smart home, and industrial applications.
These ultra-low-power wireless SoCs feature the integrated AI/ML accelerators, support for OpenThread, Zigbee, Matter, Bluetooth mesh, Bluetooth Low Energy, the highest level of industry security certification, proprietary and multi-protocol operation, ultra-low power capabilities, as well as the largest capacity of memory and flash in the Silicon Labs portfolio.
In addition, the new software toolkit was designed to enable developers to rapidly build and deploy AI and ML algorithms through the usage of tool suites including TensorFlow.
For the record, the MG24 and BG24 SoCs will be able to alleviate steep penalties associated with performance and energy use, as they are the first ultra-low powered devices with built-in AI/ML accelerators.
This specialized hardware is designed to efficiently and rapidly handle complex calculations, with internal testing ensuring a 4 times increase in performance improvement, along with over a 6 times improvement in energy efficiency.
Moreover, Silicon Labs partnered with leading AI and ML tools providers such as Edge Impulse and SensiML, to help developers gain access to the end-to-end toolchain that can simplify the development of ML models optimized for the embedded deployment of wireless applications.
Through the adoption of this AI/ML toolchain with Silicon Labs's MG24 and BG24 families of SoCs and Simplicity Studio, developers can draw information from connected devices to make intelligent ML-driven decisions.
Notably, the single-die MG24 and BG24 SoCs combine a 78-MHz ARM Cortex-M33 processor, industry-leading 20-bit ADC, high-performance 2.4-GHz radio, an optimized combination of RAM (up to 256 kB) and Flash (up to 1536 kB), and AI/ML hardware accelerator to process ML algorithms.
Source credit: