Scaling Ultra-Low-Power Edge Intelligence For Smart Devices

For decades, the data collection pipeline for sensors has been the exact same—measure, transmit, and process elsewhere. While it’s been a failproof method all these years, it’s also resulted in a large amount of energy consumption, meaning your smart watch could have a longer battery life.

Neuronova is aiming to change things up. The company’s goal? Empowering the next generation of smart devices while doing more with less. They’re bringing intelligence to the forefront and embedding it directly into the sensors. By doing this, smart devices can have a longer battery life and more functionality.

AI systems excel at processing large volumes of static data, but sensors continuously capture information in real time—in other words, temporal data. And that’s where today’s hardware falls short. A new class of chips is required, something built for processing sensor data at the source with architectures designed to understand time.

Neuronova’s proprietary neuromorphic chip mimics the human brain’s structure and function. It combines in-memory computing with the physical implementation of spiking neurons and synapses using analog circuits. This approach allows them to implement full networks in hardware, with the natural ability to process temporal data in real time and with extremely low energy consumption.

Moreover, Neuronova uses a fully analog pipeline that only sends information to the microcontroller unit (MCU) when necessary. In comparison, an analog-to-digital conversion (ADC) flow is more expensive for low-power systems.

To design their circuitry for reliability, Neuronova utilizes Cadence tools, from schematics to the very signoff. Spectre X Simulator is used for its gold standard in simulation, Genus Synthesis Solution and Innovus Implementation System are used for synthesis and place and route, and Pegasus Verification System is used for DRC, LVS, and the signoff before the chip is sent to the foundry.

Neuronova’s low-power chips make it ideal for edge AI, IoT, robotics, and more. To make things futureproof, they’re working on sensor integration to bring their chip into the package of every sensor. With the help of Cadence, Neuronova is providing a new generation of sensors that are not just measuring the world but perceiving the world with intelligence. Learn more about how Neuronova is scaling ultra low-power edge intelligence.