NN-based implementation on edge devices
The main challenge for edge-level computation is the available processing power and memory. Ignitarium has engineered the right balance by figuring out a suitable neural network (NN) architecture that can be implemented on microcontroller devices. Most of these devices have RAM under 30 KB. Ignitarium has implemented its solution to suit the limited processing power and memory in these devices. As it is one of the most popular series of processors in this category, Ignitarium chose to implement its solution on the Arm Cortex-M4.
Ignitarium supports practical audio AI solutions, such as noise suppression, on low-cost edge devices and is working closely with the vast Arm AI ecosystem to identify scenarios where the intelligent combination of deep learning and DSP brings the best of both worlds on low-powered microcontrollers deployable on various edge devices. Ignitarium chose the Arm Cortex-M platform because of its strong ecosystem, power efficiency, and mature software support for advanced math and deep-learning functions (for example, using CMSIS-NN).
Discover the Real-World Impact of Edge AI
Explore in-depth use cases that show how edge AI is powering the next generation of IoT, solving real-world problems, driving faster decisions, and lowering costs through smarter operations right at the device level.