EIKO’s platform is a three‑layer multi‑radio mesh with edge AI on every node. Federated learning lets the network improve over time without central dependence, and cryptographically‑secured telemetry keeps operators confident in every decision.
Lightweight and power‑efficient, Relay Nodes push coverage deeper into harsh environments. They form the resilient backbone that keeps control traffic flowing even when conditions change or equipment is damaged.
Typical ranges (LoRa 868 MHz): up to 10–15 km line‑of‑sight outdoors, commonly 1–2 km through rock or dense structures, and 200–500 m in heavy urban clutter. Actual performance varies with antenna placement, terrain, and noise floor.
Relays continuously exchange link‑quality metrics so the AI can select the most reliable paths and adapt quickly when RF conditions shift.
Sensor Nodes reinforce the mesh and provide rich environmental and structural insight—gas detection, particulate matter, temperature and humidity, vibration/acceleration, and optional LiDAR or UWB ranging.
Data is pre‑processed at the edge to reduce bandwidth and latency. Alerts (e.g., abnormal CO₂ trends, particulate spikes, or vibration signatures consistent with structural stress) are prioritised by the QoS engine and sent along the most reliable links first.
Integration is standards‑aligned (e.g., OPC‑UA for SCADA, MQTT telemetry) to fit into existing operations without rip‑and‑replace.
EdgeAI Nodes host accelerated inference and local decision‑making (e.g., TensorFlow Lite). They execute safety policies in milliseconds—triggering evacuation alerts, isolating power, or reversing ventilation when thresholds are exceeded—and they can reconfigure routing to preserve the mesh during faults.
Federated learning lets nodes train locally on their environment and then share anonymised updates. The system converges toward higher accuracy without exporting raw data off‑site, improving privacy and reducing backhaul load.
All actions are signed and auditable; operators can trace “why” a decision was made, node by node.
The AI continuously chooses the best medium per flow—safety control, telemetry, or bulk data—so the mesh keeps working under load, interference, or partial failure.