| Basic ideas |
| |
| The Internet of Things consists of 'things' such as sensors, devices, machines, and intelligent edge nodes of all |
| types. These things have existed at the edge since the earliest days of the Internet and have becoming increasingly |
| popular in recent times. These 'things' exist in both the consumer space and the industrial application spaces. |
| Let's call these 'things' points. These points are able to generate data with a volume, variety and velocity that |
| surpasses anything we have seen from human input. An IOT platform is one that embraces the diversity of these |
| points at the edge and facilitating their orchestration as a coherent realtime dataflow network. |
| Thus enabling action based on their collective input. |
| |
| Examples |
| 1) A large building with embedded sensors that efficiently manages the consumption of energy resources |
| 2) A complex machine such as a car or aircraft |
| 3) A home |
| 4) An industrial manufacturing plan |
| 5) A complex network of inner city traffic controllers |
| |
| What are the basic requirements of an IOT platform that would enable the broad vision given above? Lets begin by |
| looking at the points on the edge |
| |
| Points |
| 1) Able to deal with a variety of protocols both legacy and modern (e.g. Bacnet, Modpus, IP, etc) |
| 2) Able to translate the data into a standard format from any protocol (e.g. id, timestamp, value) |
| 3) Able to deal with the velocity of real-time data on ingress and egress |
| |
| Computing at the Edge |
| 1) Able to intelligently filter data at the edge |
| 2) Able to intelligently transform data at the edge |
| 3) Able to intelligently control points on the Edge |
| 4) Able to operate safely when disconnected from the network (important if you are controlling expensive equipment) |
| 5) Able to detect errors on the Edge |
| 6) Able to restart failed filters, transformers and standalone processes (Observation and Supervision of edge processes) |
| |
| Dataflow Networks |
| 1) Able to create, update and delete dataflows that define a solution |
| 2) Able to define rules that enable decisions in dataflows |
| 3) Able to define actions that are called based on decision rules |
| 4) Able to monitor actions and rules and restart flows if any of these fail |
| 5) Able to define a broad class of actions that can be executed if a decision is made via set of rules |
| 6) Ability to run dataflow networks at different levels of the IOT "hierarchy" - in devices, in connected |
| standalone servers or in clusters. |
