Ablny Technologies AS · EPCIC Edge AI · Part 1 of 2

Edge Device & Gateway Layer
Live Process Stream

Streaming Live
AlexNet-Edge Active
Split Inference: Standby
Confidence: 94.2%
Animation Speed
Edge Device (EPC Site) Edge Gateway / Local Server ⬡ Sensor Raw Telemetry ▽ Preprocessor Normalize · Filter ◉ AlexNet-Edge Quantized CNN ⚙ Inference Engine Decision Logic ⊞ Local Storage Encrypted 🔒 TPM Module Trusted Boot 🔐 Edge Security Agent Encryption · Auth △ Azure IoT Edge Identity · Routing ◈ Azure Arc Edge Mgmt ⊕ Aggregator Multi-Device Fusion ↻ Model Updater Signed Deploy ⋈ Offload Controller Threshold Logic ⋊⋉ Split Inference Manager ⬡ Secure Proxy TLS · Auth → Azure Cloud (Part 2)
Active Pipeline Step
Sensor Capture
Preprocessing
AlexNet CNN Inference
Confidence Evaluation
Local Storage · TPM Attest
Security Agent Sign
Azure IoT Edge Route
Gateway Aggregation
Split Inference (if needed)
Secure Proxy → Cloud
Live Metrics
Inference Latency11ms
Confidence Score94.2%
Frames Processed0
Cloud Offloads0
TPM Attestations0
Anomalies Detected0
Event Log
Solution: EPCIC Edge AI
Architecture: Part 1 of 2 — Edge & Gateway
Cloud: Azure IoT Edge + Arc
Model: AlexNet Quantized INT8
Security: TPM 2.0 · mTLS · Zero Trust
Part 2: Azure Cloud Full Inference →