Skip to main content
STUB DOCUMENT: This page is intentionally minimal and will be expanded with deeper technical details in a future update.
When an edge device has multiple cameras (each represented by a separate digital twin), Edge Core starts one driver container per camera and a single shared worker container that receives frames from all of them.

​
Architecture

┌──────────────────┐   ┌──────────────────┐
│ cyberwave-driver  │   │ cyberwave-driver  │
│ (Camera A)        │   │ (Camera B)        │
│ TWIN_UUID=aaa...  │   │ TWIN_UUID=bbb...  │
└────────┬─────────┘   └────────┬─────────┘
         │ Zenoh frames/default  │
         ▼                       ▼
┌─────────────────────────────────────────┐
│         cyberwave-worker-{env[:8]}       │
│  TWIN_UUIDS=aaa...,bbb...               │
│  Subscribes to both camera channels      │
└─────────────────────────────────────────┘

​
Driver startup sequencing

Edge Core starts drivers before the worker and verifies each driver container reaches a running state:
  1. All linked twins are resolved from the environment.
  2. Driver images are pulled in parallel.
  3. Containers are started sequentially, each verified with a readiness probe.
  4. Once all drivers reach a stable state, the worker container starts.
  5. If some drivers fail, the worker still starts (with a warning) so healthy cameras can be utilized.

​
Worker environment wiring

The worker container receives:
VariableValuePurpose
CYBERWAVE_TWIN_UUIDSComma-separated list of all linked twin UUIDsWorker subscribes to all twin channels
CYBERWAVE_TWIN_UUIDFirst twin UUID (backward compatibility)Legacy single-twin workers

​
Model pre-download

Before the worker starts, Edge Core scans workers/*.py for cw.models.load("...") calls and pre-downloads all referenced models to the local cache (~/.cyberwave/models/).

​
CLI visibility

cyberwave edge status shows:
  • Each driver container with its twin UUID mapping
  • The worker container with its full twin list
  • Running/stopped status for all containers

​
Health monitoring

Edge Core continuously monitors driver containers during the runtime loop:
  • Restart loop detection: Drivers that restart too frequently are stopped automatically.
  • Driver-down alerts: If a driver stops while the worker is running, an alert is sent to the associated twin.
  • Worker health: The worker container has its own health monitor with circuit-breaker protection against crash loops.