Every sensor vendor, every city department, every data feed speaks a different format. CSV, JSON, GeoJSON, GTFS, GBFS - all incompatible. Without CASSIOPEIA, every new data source means weeks of custom integration code. 

CASSIOPEIA maps any source to NGSI-LD Smart Data Models - once. No code. Repeatable. Validated.

• Transforms multiple input formats: CSV, JSON, GeoJSON, GBFS, GTFS, and others.
• Outputs NGSI-LD entities in different representation modes (normalized/concise/simplified).
• Advanced mapping configuration: templates, conditional logic, nested objects, relationships, list relationships.
• Synthetic entities: create referenced entities on-the-fly when source data only contains partial references.
• Smart Data Model integration: download/search/list schemas and validate outputs when schemas exist.
• Flexible writers: write to files (default) or push directly to a Context Broker via HTTP.

• Standard feed shortcuts: pre-built configurations for common GTFS and GBFS entity types.

• Not a Context Broker - it generates/publishes entities to one. 
• Not a data catalog / data space - it’s a transformation/mapping component in a larger stack. 
• Not a full ETL orchestration platform - it focuses on mapping + validation + output (scheduling/orchestration belongs elsewhere). 
• Not an analytics/ML module - no mention of modeling or KPI logic, just transformation. 

• Not a UI-first product - usage is documented mainly via CLI + config files. 
  

• Inputs
• • Source files or feeds: CSV / JSON / GeoJSON / GBFS / GTFS.
  • Mapping config files: JSON5 (recommended) or JSON.
• Processing
• • Applies mapping rules using Tera templates with support for conditionals, filters, and nested navigation.
  • Generates entity IDs using dataModel + entityName + optional URN parts (site/service/group).
  • Performs type transformations (string/int/float/bool/datetime/geo types).
  • Optionally validates against Smart Data Model schemas (downloaded via Cassiopeia download).
• Outputs
• • NGSI-LD entities written as files, or pushed to a Context Broker (--writer-type context-broker --broker-url ...).
  • Output representation modes: normalized, concise (default), simplified.

• • Null handling: include or skip null values.

Cassiopeia sits between “raw data sources” and your NGSI-LD runtime:

• Upstream: files, APIs, feeds, exports (CSV/JSON/GeoJSON/GTFS/GBFS).
• Cassiopeia: mapping configs + optional validation against Smart Data Models.

• Downstream: file outputs (for pipelines/tests) or a NGSI-LD Context Broker (for live platforms).

• Language: Rust
• Config format: JSON5/JSON mapping configs
• Templating: Tera template engine
• Interfaces: CLI (cassiopeia ...) + server mode (cassiopeia server)
• Outputs: File writer or Context Broker writer (HTTP)

• Validation dependencies: Smart Data Model schemas downloaded by tool command

Cassiopeia provides five interfaces for different stages of the data integration workflow:

• CLI - Command line -  Batch jobs, CI/CD, one-off conversions
• Server -   REST + WebSocket -  Centralized hub, live data streams
• Explorer -  Terminal (TUI) -  Schema discovery, model inspection
• Wizard -  Terminal (TUI) -  Interactive mapping creation
• Configurator -   Web browser -  Same as Wizard, graphical UI 
  

FIWAREBox Cassiopeia -  Data Model Selection

FIWAREBox Cassiopeia - Attribute Mapping

• standalone/self-hosted
• embedded in FIWAREBox
• managed as part of FIWAREBox (NGSI-LD Context DataHub)
  

TBD

IT personas: architects, data engineers, integrators, pilots/research  

Auriga, Lynx, Libra