IoT Data Models

A useful IoT data model represents deployments (sensors/actuators), environments (rooms/locations), domain concepts (users/activities), readings, and context/situations. It should support interoperability, integration with other data sources such as the Web, multiple domains, and efficient management.

The lecture surveys key-value, markup, graphical, object-oriented, logic-based, ontology, and linked-data models.

Key-value models

A key-value model stores simple pairs such as (temperature, 20). It is easy to manage and useful for simple context/service matching, but the value is largely opaque. Relationships between entities are not first-class and sophisticated retrieval may require application-side parsing.

Linked/ontology-based models

A linked-data model represents entities and typed relationships as a graph. For example:

Sensor S --observes--> TemperatureReading R
Sensor S --locatedAt--> Room A
Reading R --hasUnit--> Celsius
Room A --partOf--> Building B

This can answer relationship questions such as “which sensors in Building B measure temperature?” without treating the whole value as an opaque blob. Semantic Sensor Network (SSN), SensorML, O-DF, and IPSO Smart Objects are examples/standards discussed in the material. O-DF can serialize objects, info items, values, and timestamps in XML or JSON-like form.

Exam contrast

A key-value model can store a shipping container and its owner/location as fields, but it does not naturally represent/query a network of linked container, vessel, port, owner, and event entities. A linked model makes those relationships explicit and interoperable. Do not claim key-value is unable to store nested JSON; the important limitation is that relationships are not modeled/queryable as first-class graph links.

Sources

• master-cs/raw/iot/4.2-IoT-Connected Data - DataModels.md
• master-cs/raw/iot/Question Sheet Week 5.md