There has been talk of the Internet of Things and Industrial IoT for quite some time now and it is starting to take shape. A wide variety of sensors, (industrial) routers and cloud-based dashboards are created.
There are 2 architecture directions:
1. Fully cloud based
Data loggers, which send data via CDMA / GSM or eg LORA directly to an external server. Applications include smart electricity meters.
2. Systems with partly Cloud and partly local activity / intelligence
The router has a fairly central position in the infrastructure. This serves as a node for various types of connectivity from sensors and to actuators or to local systems / networks. A local PLC system is an example of this for the industry.
The architecture with local activities is called EDGE COMPUTING
Edge computing offers several advantages over cloud solutions:
• Faster response (low latency), it limits the required internet bandwidth (which is becoming a bigger problem)
• Remains within company network (security)
• Keep on running even if there is no internet connection for a while
Edge Functionalities: actually very middleware layer between hardware / sensors and the cloud. So including router, IoT gateway and local PC / laptop / tablet and raspberryPI.
Machine Learning can also run on the edge, the more computing power there is. An example of this is the self-driving car, in particular latency and network bandwidth are key. In addition, a local process must continue to be able to continue, even if communication with the cloud is lost.