BrainChip Holdings (ASX:BRN) launched a neuromorphic computing reference platform built on its Akida AKD1500 processor to enable real-time radio frequency (RF) signal classification for defense and edge applications under strict power and size constraints, according to a Tuesday statement by the company after market hours.

The platform was designed for defense, intelligence, and government applications, enabling on-device signal intelligence in constrained environments where limitations in power, size, weight, and connectivity make graphics processing unit or field-programmable gate array-based solutions impractical.

The company said the platform delivers sub-watt inference while classifying over 20 RF modulation types in real time at over 85% accuracy and adapting to new waveforms through data capture for retraining.

The system integrates a software-defined radio and serves as a reference design for prototyping battery-powered, edge-deployed signals intelligence applications, the company added.