Brain-Inspired "NeuroRadar" Delivers a Claimed 97 Percent Energy Saving for Smart IoT Devices

Researchers at the University of California San Diego and the University of Virginia have published a paper detailing a brain-inspired neuromorphic radar sensor for ultra-low-power Internet of Things (IoT) systems — ideally those integrating hardware for processing spiking neural networks (SNNs).

"NeuroRadar [is] a novel low-power radar-sensing system that fully exploits the power of neuromorphic sensing and computing," the researchers claim of their creation. "NeuroRadar draws inspiration from neuromorphic sensors that mimic mammalian sensory systems, generating event-triggered outputs in response to external stimuli. Contrary to traditional radars with continuous frame-based outputs, NeuroRadar produces spiking patterns upon detecting motion in the surrounding area."

Compact, low-power radar systems are used for everything from simultaneous localization and mapping (SLAM) in autonomous vehicles to gesture control in smartphones and wearables, but there's a limit to how small and energy-efficient they can get thanks to the need for a relatively complicated radio-frequency front-end. NeuroRadar, by contrast, is designed around an array of self-injection locking (SIL) sensors — reducing the power-hungry active components down to just a low-power free-running oscillator.

"NeuroRadar converts ambient motion signals from the sensor front end into spikes using an analog spike-encoding circuit," the team explains of the sensor's operation. "The spike encoder follows a biological neuron model, preserving all the essential sensing information in the spike sequences. The spike sequences can then be directly processed by the SNNs on neuromorphic computing systems, thereby eliminating the need for any non-spike-based computing units.

"Consequently," the researchers continue, "we can train the SNNs using these raw spike signals for various tasks, including gesture recognition and localization. This comprehensive SNN processing workflow allows NeuroRadar to deliver application-specific sensing results with superior energy efficiency."

Compared with a rival multi-tone Doppler radar system. Doorpler, with conventional RF front-end, the researchers claim, NeuroRadar delivered a larger coverage area despite 10dB lower transmission power along with "one to two orders of magnitude" reduction in front-end power. "Unlike Doorpler, which merely detects crossing events and their direction, NeuroRadar offers both location and speed estimation," the team continues. "At the same time, SNN processing significantly reduces the computational power, and the end-to-end system power consumption is reduced by 97 percent."

The team's work is published in the Communications of the ACM under open-access terms; a supporting technical perspective is available on the CACM website.