February 28, 2017 – The implementation characteristics of a Smart Manufacturing solution depend on numerous strategic, economic, and technological factors.
The architectural principles of this can be summarized in one exemplary scheme, like that shown below, which seeks to include all possible exceptions while also depicting the typical elements constituting the Smart Manufacturing architecture.
Each practical implementation can be based on a subset of the elements described, as well as being a function of developmental maturity of the specific project.
The paradigm of Cloud Computing goes well with the functional needs of Smart Manufacturing: in fact, it is easy to understand how the pooling of centralized intelligence can ensure maximum efficiency and flexibility in collecting and processing data from a potentially large set of distributed sources.
It’s realistic to think that we can refer to a Private Cloud model for the management of data and the internal processes of Manufacturing companies for which the maximum levels of security and privacy is necessary. Such architecture can indeed be open to integration with Public Cloud components to enable the acquisition of data outside the enterprise perimeter (such as environmental sensors, data collected by partners or public institutions, etc.).
In broad terms, the architecture related to IoT solutions as applied to the industry (at least if we refer to the Private Cloud model) can be organized into seven hierarchical levels, ranging from “objects” physically present in the field to applications that manipulate, analyze and present the collected data.
This hierarchy can correspond to the transformation of a mass of (raw) data into synthesized, valuable information, which in turn could be brought to knowledge and awareness:
The physical layer of the “things” that generate data include virtually any type of Smart object, that which is endowed with intelligence and the ability to communicate with the outside. This ranges from sensors to PLCs (Programmable Logic Controllers), counters, video cameras, HMIs (Human-Machine Interface), smartphones, RFID tags, SCADA servers, etc.
The Field Area Network (FAN) is a network device which allows the collection of data generated by different objects. By nature, this is a network as heterogeneous as the objects with which it must interface, based on a potentially large number of protocols and standards: Zigbee, Bluetooth, Wi-Fi, WirelessHart, RS482, etc. Typically a gateway performs conversions from the protocols and access technologies to the IP (Internet Protocol), the de facto standard for data communications.
We’ve already talked about the importance of Cloud logistics, however, there’s no point in conveying the entirety of the immense amount of data collected at the periphery to the center. Here, an Edge Computing layer becomes strategic; this is a local intelligence component allowing the performance of a first treatment of data by the filtering and implementation of application logistics at a local level. The ultimate goal is to increase reactive speed in the face of events that require immediate action (erasing the latency inherent in the Cloud architecture) while reducing redundant information sent to the central server.
A modern IP network can become an appropriate direct connection to industrial installations, with the IT infrastructure hosting applications of interest to the company. IP devices must be compliant with all the structural requirements in order to be placed in environments with challenging characteristics in terms of temperature, humidity, vibration, interferences, etc. Wireless solutions allow us to offer extensive network coverage while reducing wiring costs. A DMZ (De-Militarized Zone) made through the use of firewalls allows the logical separation of the industrial network from that of the office, maximizing security levels. Finally, latest generation IP devices can also be designed to internally host the logics of Edge Computing.
A new generation Data Center lies at the heart of Private Cloud infrastructure dedicated to the collection and processing of data. The components of Computing, Storage and Networking must be optimally engineered and coordinated for maximum efficiency and scalability. The Data Center is the physical end of the data collected from the objects and transferred throughout the network; the data is stored in a Database and/or File System in the Data Center.
The Data Abstraction Layer basically acts as an interface between the overlying application layer and various Databases and File Systems in which data can be saved. Its function is to aggregate and logically tailor information according to the needs and expectations of specific applications irrespective of the method and the formats used to save and store the data (SQL-based DB, flat file, big data store, web services, excel files, etc.).
The application layer is that where the actual processing of the information occurs. Applications are extremely diverse, including elements as varied as Analytics Data, Business Intelligence, localization software, Energy Efficiency software, predictive maintenance, video stream management, root-cause analysis systems and lastly, real industrial level applications such as MES (Manufacturing Execution System), MRP (Manufacturing Resource Planning), etc. This layer is therefore about the extraction of valuable content from raw data to be used to implement business logistics using the best information which arrives in real-time from production plants.
Fully parallel to the Private Cloud architecture, Public Cloud infrastructure can also allow the bringing in of application-level data collected from the field. In this case, it can be assumed that this is either data external to a single manufacturing company, where the same stringent confidentiality requirements related to internal industrial production data do not apply, or data collected from heterogeneous sources and distributed for noneconomic integration with Private Cloud architecture. In this context, information collected by the objects through the access network (FAN) are transferred on Public Cloud directly through a public WAN (e.g.: 3G, LTE, WAN-Low Power).
The overall architecture can therefore be described as a Hybrid Cloud model in which the various applications are linked to enable the integration and sharing of information. Of course, the integration of different Private Clouds can also be helpful in managing information at the level of integrated supply chains.
Discover more about Italtel’s IoT offer line.