A rose by any other name… Part 2

Part 1 of this blog explained how systems referred to as the Internet of Things, Industrial Internet, and Smarter Planet are all based on machine-to-machine (M2M) technology and essentially very similar in nature. Each of these systems makes amazing applications possible, by receiving information from, and controlling remote assets, device or things. But how does a ‘centralized’ application interact with, for example, a ‘remote’ wind turbine or a traffic light?

Part 2 of this blog will describe at a very high level the architecture these systems share, and the key components underlying these systems. After reading this, you will understand the function of these key components in the overall system.

Consider Figure 1, which shows a central intelligent application controlling a remote wind turbine (i.e. the Industrial Internet) and a remote traffic light (i.e. Smarter Planet).

Figure 1

These key components are shown in Figure 2 depicted in a chain.

Figure 2

The asset, device or thing (1) that is being monitored and/or controlled (i.e. the wind turbine and traffic light of Fig 1) will typically be located remotely from the control center. As previously discussed, there is no limit to what these assets might be – and within a few years they are expected to touch every aspect of our lives, from mundane household items to high end industrial equipment and civil engineering structures.

The remote device will be equipped with some kind of sensor (2). The sensor will measure one or more parameter(s) associated with the device, such as a pressure reading, temperature, location, etc. These sensors are typically solid state, and rapidly declining in price as production volumes increase. This will only accelerate their ubiquitous deployment.

The sensor is connected to a modem (3). The modem is responsible for sending the sensor data to the network. The exact modem technology will depend on the network employed. Sometimes this will require a regular wire-line modem – assuming a direct connection to the Internet is available (per the traffic light example); at other times a wireless modem (per the wind turbine example). In some instances it will make sense to combine the sensor and the modem into a single unit.

The network (4) is often in reality several networks. Almost all these systems make use of public or private Internet.  Providing network coverage for the ‘last mile’ to the device or thing is often where wireless networks come into play. Today’s cellular networks (2.5G, 3G and 4G) operate as true extensions of the Internet. In other circumstances, it can be far more economical and efficient to use other network technologies including Wi-Fi, Zigbee and Bluetooth to name a few.  Often a local network is used to collect the sensor/modem data and feed it into a cellular gateway to connect to the Internet.

Once the data traverses the Internet, it arrives at a data store (5).  With thousands or even millions of devices in the field generating structured and unstructured data, big data provides an ideal repository for this data.  Over the past few years, the use of big data has exploded in popularity and led to highly distributed deployments that offer the dual benefits of high resiliency and security.

How do we make sense of all the masses of collected data? This is the role of specialized software known as analytic software (6), which can rapidly sift through the data and correlate significant events. Here lies a great deal of the power of these systems. An element of data –say a bearing pressure reading on the wind turbine – standing by itself might be of little interest, whereas the trend of a set of pressure readings taken over a certain time interval may be used to predict an impending failure. This is incredibly useful information and represents actionable knowledge, i.e. shut down the wind turbine before the pressure reaches a level that would cause a very costly failure of the turbine. In this way, analytic software working together with big data can convert an incoherent mass of raw data into an actionable insight.

The corrective action is taken by intelligent control software (7), which is the ‘brain’ of the system. The control software receives significant and actionable information from the analytic software. Based on this information, the control software will send control signals in the outbound direction to change the status of the remote asset. The control software will also likely make a data entry in the big data store (creating an audit trail of changes) and then traverse the network(s) and modem to reach the sensor/ asset in the field to effect the change.

One other critical consideration is security. Each element in the architecture will feature some level of security capability to ensure that the data collected stays within the system, and that the remote devices are protected from unauthorized malicious access.

This article has presented a very high level overview of the typical architecture employed in the Internet of Thing/Industrial Internet/ Smarter Planet type systems. While each real-world system will have its own unique features, you will now be able to identify some of the key elements that will appear in these systems and understand their main function.

Leave a Reply

Your email address will not be published. Required fields are marked *