The complete NeuroNAS system uses a four-chip, three-zone layered architecture. Capture, transmission, inference, and control each run as independent modules that work together seamlessly, with no reliance on the cloud at any stage. The following are target specs; final production parameters are subject to launch.
| Wireless SoC | nRF5340 · BLE 5.4 + Thread |
| EEG Coprocessor | GAP9 · denoising / FFT |
| Function | Real-time multi-channel EEG reception, filtering, and feature preprocessing |
| Neuromorphic Chip | BrainChip AKD1500 |
| Compute | 800 TOPS |
| Power | < 300 mW · event-driven |
| Deployment | Direct ONNX deployment |
| Main SoC | RK3588 · 8-core ARM + 6 TOPS NPU |
| Memory | 8GB – 32GB RAM |
| Storage | 1TB – 16TB NVMe (encrypted · hot/cold tiering) |
| Local Gateway | Built-in Matter 1.4 gateway |
| Routing | Thread border router |
| Bridge | Zigbee bridge · cross-brand interoperability |
| Parameter | KidiFlow Headband | DreamSync Sleep System | BCI Controller |
|---|---|---|---|
| EEG Channels | 8–16 channels | 4–8 channels | 2–4 channels |
| Electrodes | Graphene dry electrodes | Behind-ear dry electrodes | Head EEG dry electrodes |
| Amplification / Sampling | Low-noise amplifier · 12–24-bit ADC | Low-noise amplifier · 12–24-bit ADC | Low-noise amplifier |
| Wireless | BLE 5.4 | BLE 5.4 | BLE 5.4 |
| On-Device Compute | Zero compute · sensing only | Zero compute · sensing only | Zero compute · sensing only |
| Core Rhythms | theta / beta focus rhythms | Delta / SWM slow waves | Motor imagery ERP |