Abstract: A method for processing of a data stream using a graph-based data model includes receiving a data stream including data messages; disassembling the data messages data elements and metadata; generating a structured data model comprising the set of data elements based on the type of the data elements and the pattern of the data messages; instantiating a workflow to process the structured data model; configuring a CPS-G model sub-graph to add to the CPS-G model based on the type of the data elements and the pattern of the data messages; adding the CPS-G model sub-graph to the CPS-G model to form the CPS-G dataset; and storing the CPS-G dataset including the CPS-G model sub-graph in a data store for further processing by streaming computation and machine learning algorithms.

Abstract: A configurable data collection, processing and delivery system implements a unified data-flow execution pipeline in collection, processing and delivery of data to a target within or external to a computing device. Multiple data collection sets are defined from data available to the system and organized into at least one tensor based telemetry data stream associated with a service flow. At least one dynamically modifiable execution unit is provided to process the telemetry stream and the service flow. The telemetry stream is also processed into one or more target destination delivery streams that adheres to protocol, network characteristics, and transmission process required by the set of targets.