DESCRIPTION
Nowadays, every part in the production process can be interconnected – starting from the sensor up to the most complex machine. This is made possible with the use of standard protocols, giving us the ability to interconnect the “smart factory”.
Interconnection requires any process actor to be interoperable and to have the ability to interact and work with any other actors (installed or not yet so) and without restrictions. In the past, the request for a product was to be “compatible” with another one. That meant a designed ability to be connected to another product by means of a proprietary standard. Not long after that, the request was to be compatible with some de facto standard, which was typically the standard fixed by the market leader. Nowadays, the request is that the product must use and implement open standards fixed by committees, or foundations that are universally accepted.
The adoption of open standards cannot only be limited to the connection issue, but must also be used to guarantee the interoperability of storage and elaboration. Only in this way, we obtain additional value from any new generated data.
Traditional tree topologies are now outdated by the usage of new kinds of links and networks. These links and networks, with completely connected meshes, remove the dependency from the specifications imposed by the producer of a specific node. Node resources are no longer passively queried, but are active, with the ability to send data through the network and to the Internet. This is done by using patterns like Publish & Subscribe; so doing, data producers and data consumers are decoupled and more consumers can use the same data to implement one's own "business logic.”
The existence of IIoT requires the adoption of a new data model:
- Application: it implements the Thing's "business logic"
- Thing: "stateful" resources are uniquely identified by a URI (Uniform Resource Identifier), that is a unique global resource identifier
- Transfer: standard protocols (REST, Pub-sub, HTPP, AMQP, MQTT) must be used
- Transport: UDP, TCP protocols
- Network: Ethernet, Wi-Fi, WiMax, 4/5G Mobile networks
A continuous growth in production of data requires new storage techniques, new analysis algorithms, and distributed computing. This means that more expertise and expensive infrastructures are needed to keep up with the industry. On the other side, the growth in bandwidth availability accompanied with its dropping cost, suggest a shift in storage, analysis, and computing to the Cloud should be made. Switching to applications, like the Cloud, allows a company to use services implemented by third-parties, who offer specialized, always updated, and reliable solutions. Big Data Storage, Data Mining, Data Analytics, Machine Learning, and Artificial Intelligence can now be implemented using consolidated solutions, while the company can commit to its core business.
