A new AI-enabled tech from Australia-based firm BrainChip could classify radio-frequency (RF) signals in real time under one watt of power.
Powered by an AKD1500 chip processor, the Akida Communication Reference Platform is designed to deliver continuous signal inference while using significantly less power.
It supports signals intelligence (SIGINT) missions without the power, thermal, and size constraints associated with conventional edge AI modules.
Designed for portable deployments, the platform integrates with software-defined radios such as the USRP B205mini and Epiq Sidekiq through a Raspberry Pi 5 host, allowing operators to perform RF signal classification without wired power, bulky hardware, or cloud connectivity.
Engineering teams can also use the system to develop custom intelligence, surveillance, and reconnaissance (ISR) capabilities and other SIGINT applications.
“BrainChip’s Akida Communication Reference Platform proves that real-time SIGINT can be condensed into a portable battery-powered solution to extend the range of deployment options,” said Sean Hehir, Chief Executive Officer of BrainChip.
“This extends BrainChip’s reference platform strategy… to demonstrate that Akida is a broadly applicable edge AI processing engine for many use cases.”
AI-Enabled Classification
The device can also be used as a hardware reference design kit for evaluating or prototyping neuromorphic AI-based RF signal analysis.
According to the company, it can recognize more than 20 modulation types with more than 85 percent accuracy at a 30-decibel signal-to-noise ratio.
By applying neuromorphic AI, the system captures previously unrecognized waveforms from emerging emitter threats, allowing models to be retrained and adapted to recognize new signals.
This reportedly helps address a limitation of traditional digital signal processing classifiers, which are often rigid, power-hungry, and too large for portable deployment.
