Introducing “IoT Platforms”: What is IoT?
Remote device access is not new in factory automation. However, that access has been dramatically changed lately, enough to change how future factory systems will be designed, installed, managed and used. “Traditional machine-to-machine [M2M] solutions typically rely on point-to-point communications using embedded hardware modules and either cellular or wired networks,” says Chantal Polsonetti, vice president, ARC Advisory Group. “In contrast, Internet of Things [IoT] solutions rely on Internet protocol [IP]-based networks to interface device data to a cloud or middleware platform.” That is, all devices–actuators, machines, sensors, controllers and so on–will one day be assigned IP addresses.
Integration of device and sensor data with big data, analytics and other enterprise applications is a core concept behind the emerging IoT.
But there’s more, Polsonetti continues. “Though M2M solutions offer remote access to machine data, these data are traditionally targeted at point solutions in service management applications. Rarely, if ever, are the data integrated with enterprise applications to help improve overall business performance. Integration of device and sensor data with big data, analytics and other enterprise applications is a core concept behind the emerging IoT.”
How different is this? Says Matt Hatton, founder and CEO of Machina Research, “M2M is the equivalent of broadband; IoT, of Facebook and Google.”
End the brittleness with new technology It used to be, according to Russ Fadel, president of ThingWorx, that the data communications found in, say, factory environments “consisted primarily of sensors, devices and equipment; communications elements; general purpose development tools and fixed applications, all trying to work together through a common operating network. However, they typically consisted of ‘rigid packaged applications' or ‘brittle custom applications.’”
These applications typically required complex programming, tailored to specific projects, were difficult to maintain and evolve, risky and costly, and “essentially created barriers to future innovation.”
That all changes in the industrial IoT (IIoT; “Industry 4.0” in Europe). First, there’s the change from simple devices that respond to and transmit on/off signals, to smart devices that sense and respond to numerous aspects of an operation and its operating environment, and then spew out torrents of data about both. “In streaming, there’s data that’s always in motion. It’s incessant,” says Santosh Balasubramanian, Microsoft Azure CTO. Second, there’s the change from focusing on things to focusing on the data coming from those things and analyzing the data. Last, there’s the change from data as data, to data as intelligence–intelligent things, intelligent systems, and increasing the intelligence of those using those things.
According to Hatton, “You design the application, you build the device, you put all this stuff together. But when you think about the IoT, what you have to do is mashup a whole load of sensor data with other forms of data, make it available for the applications developers to build out applications and add sophisticated interfaces. Put together a bunch of different elements.” Almost simple, yes? Not surprisingly, designing and installing IoT-based systems requires some new software tools. For example, points out Balasubramanian, developers try to make SQL and other classic databases work despite problems caused by processing a flood of streaming data. “They might not know that tools exist for them.”
At its core, the “IoT platform” is a collection of software tools for developing, deploying and maintaining IoT applications. And all IoT platforms have a common goal, says Mayank Thapa, bigdata associate at smartData Enterprises pvt ltd: “connect the data network to the sensor arrangement and provide insights using backend applications to make sense of the plethora of data generated by hundreds of sensors.” That connection may be through gateways; the data network will include cloud services; the backend applications can be anything; the plethora is another way of saying “streaming data;” and making sense will involve data analysis, visualizing real-time data, maybe augmented reality, and triggering some type of response, such as device control or “liveware” decision making.
Selecting an IoT platform
IoT platforms are relatively new; however, market analysts estimate well over 100 IoT platforms are already available worldwide. The vendors include the usual suspects: traditional IT companies (e.g., Atos, HP, IBM and Microsoft), traditional controls suppliers (e.g. Bosch, GE and Siemens), traditional consulting firms (Capgemini and CSC), and vendors of varied heritages, such as CAD/PLM (e.g., PTC), microprocessors (e.g., Intel) and telecommunications (e.g., Deutsche Telekom). There are also newcomers, “pure-play” vendors that specialize in certain aspects of IoT implementation.
As in the early days of any manufacturing technology–think CAD/CAM, MES and MRP/MRP II/ERP–consider the IoT platform and vendor based on flexibility, agility, breadth of technology offering and local support. Focus on vendors and platforms that best meet specific business needs and IT requirements. Migrating factory systems from conventional M2M to IoT does not require “rip-and-replace”; parts of an existing IT infrastructure are still valuable.
Evaluating the technology is relatively easy: Will the platform communicate with existing computer-based devices, sensors and networks; is it based on current (and known future) standards; does it meet current (and known future) capabilities for device connectivity and management, mashup creation, analytics, cloud services, data storage, data presentation and user interaction (i.e., “user experience”); and so on? Where is data stored, and how is data latency minimized during data retrieval? Don’t forget how the platform deploys data security and privacy, and to what extent.
Consider the accompanying software utilities for developing, visualizing and customizing digital models and templates that mimic the physical reality of “things.” How effective is the platform at testing and proving out new IoT applications and those applications’ relevance to a company’s business processes, profitability and future goals?
A primary piece of any IoT platform is analytics. This piece “should be able to shore up data analysis and presentation and provide effective mechanisms for data preparation, data visualization and reports generation,” says Chris Kuntz, vice president of marketing at Glassbeam, a machine data analytics company. It should also, he adds, include predictive analytics to deal with future effects on a business.
The build-versus-buy argument remains the same. “In-house solutions often face a series of challenges, and require a variety of committed resources for an extended period of time. It is inherently time-consuming and risky if not planned properly with appropriate resources needed to not just design and implement, but also to maintain and manage its lifecycle on a continuous basis,” continues Kuntz. Consider the “one-stop suppliers” that have both application development/deployment and analytics services.
As the IIoT becomes more a mundane part in factory automation, plan on IoT platforms being one of the technological toolboxes required for developing and managing manufacturing excellence.
Check out part 2 in this series, Comparing IoT Platforms
As with so many things, the devil is in the details when comparing Internet of Things (IoT) platforms. The following overview of two such platforms is hardly an apples-to-apples comparison, but it should give an indication of the breadth of such products. KEEP READING
The smart-factory world of digital manufacturing is coming, and manufacturing-technology developers are eager to help you be ready.
Company plans to achieve 15% efficiency improvement through connecting sensors—and everything in between—to the Cloud
The latest wave of manufacturing execution systems takes advantage of the Internet of Things, leading to simpler and faster implementations and truly real-time data analysis, decision-making, and problem resolution.