Device-cloud communication models

METHODS AND SCENARIOS

• event messaging: alerts, events, time series telemetry
• file upload: raw, compressed measurement data
• device twin: set parameters, read conditions, long running commands
• direct method: perform immediate action, calibrate, reboot
• file download: firmware update
• streaming: bidirectional data stream, diagnostic console

TECHNICAL PARAMETERS

• authentication method
• protocol
• size
• frequency
• qos, priority

COMMUNICATION PROTOCOLS

• HTTPS
• MQTT
• MQTT over WS
• AMQP
• AMQP over WS
• WS

PAYLOAD FORMAT

• JSON + JSON schema
• Protobuf, Flatbuf, Nanopb
• Avro, Thrift

COMPRESSION

• snappy
• deflate
• lzma
• bzip
• brotli

API DESCRIPTION

• HTTP: OpenAPI/Swagger
• AMQP, MQTT, WebSockets: AsyncAPI

DATA STORAGE

• WARM path storage: relational database, document oriented database, time-series database, mem-cached database, graph database
• COLD path storage: data lake file format
• Hierarchy: BLOB storage: each entry in the storage doesn't conform to any particular model

DEVICE MODEL (DIGITAL TWIN)

CAPABILITY MODEL

A device capability model describes the capabilities a device should implement:

• telemetry and properties it reports
• cloud properties that aren't stored on the device
• commands that can be executed

An interface describes related capabilities that are implemented by a device or digital twin. Interfaces are reusable and can be reused across different capability models.

• telemetry describes the data emitted by a device or digital twin, whether the data is a regular stream of sensor readings or an occasional error or information message.
• property describes the read-only and read-write state of a device or digital twin. For example, a device serial number may be a read-only property and the temperature set point on a thermostat may be a read-write property.
• command describes a function or operation that can be performed on a device or digital twin.

OTHER MODELS

• Anomaly detection models: ML Anomaly detection API format
• Physical environment models: Spatial intelligence graph

DEVICE EXECUTION MODEL

• Real time operating system
• Preemptive multi tasking operating system

AUTHENTICATION

ENCRYPTION

• symmetric
• asymmetric

X.509

• no need to distribute the public keys when creating key pairs.
• can maintain a list of invalidated certificates

PROVISIONING

• group enrollment: one root certificate, and any number of leaf certificates signed by this root

have many sensors configured to provide an identical format of telemetry data

• group enrollment with intermediate certificates: organise devices into enrollment groups, sign with intermediate cert

device hierarchy: flexible access control and initial configuration parameters

• individual enrollment
  Updated 02 January 2023