Datastream Management

A Datastream is the basic building block for pattern recognition used in Falkonry. A Datastream is associated with a set of Signals (that it consumes as input), a set of Entities, and a set of Assessments that convey the output of the Datastream.

A Datastream can consume a History Window of signal data and Events that lie within the window to facilitate the learning of Models.

At any time one model can be designated as Active. If the Datastream is turned on with an active model, signal data streamed into the Datastreams will produce a live output stream of assessment results.

Creating a Datastream

A Datastream can be created using Datastream UI via file upload. The file needs to has a pre-defined structure. The same option can be used to create or update an existing Datastream.

Required File Format

The file needs to be CSV formatted with fixed header information as described below:

Column Name	Use
source_name	Name of the approved signal that exists in Falkonry. Export the signals list from the UI to get this. This is a required column.
datastream_name	Name of the datastream in Falkonry. This is a required column.
entity_name	Name of the entity in Falkonry. This is a required column.
signal_name	Name of the signal in context of the datastream. This is a short name of the signal that exists in Falkonry, and it must be the same for all entities within a datastream. This is a required column.
signal_description	Description to attach to the signal in context of the datastream. This is an optional column.
[signalProperty]	Property to attach to the signal within the datastream along with the properties already set on the approved signal. Example: sendOnChange, minThreshold, maxThreshold, etc. These are optional columns.

Note

Multiple Datastreams can be created or updated via a single CSV file. A dry run can also be performed to validate the file by selecting Dry run option on the UI when uploading the file.

Example

source_name	datastream_name	entity_name	signal_name	signal_description	sendOnChange	minThreshold
machine1/current	Machine monitoring	machine1	current	output current reading	true	10
machine1/temperature	Machine monitoring	machine1	temperature	outer casing temperature	true	32
machine2/current	Machine monitoring	machine1	current	output current reading	true	10
machine2/temperature	Machine monitoring	machine1	temperature	outer casing temperature	true	30
machine1/vibration	Quality Assessment	machine1	vibration	outer casing vibration	false	8

The above example will result in two Datastreams. Machine monitoring and Quality Assessment becomes the Datastream names with the configured entity and signals. Machine monitoring Datastream will have two entities (machine1 and machine2) and two signals (current and temperature) whereas Quality Assessment Datastream will have one entity (machine1) and one signal (vibration).