The Intelligent Edge for the Internet of Things
In the Internet of Things, Look at the “Fog” between Devices and the Cloud
Much is made of Big Data streaming from small devices to huge server farms in the Cloud where it is supposedly analyzed and used for a vast array of applications. But in addition, there is a big potential for utility at the points were data is aggregated from local networks attached to the Cloud.
BY TOM WILLIAMS, EDITOR-IN-CHIEF
Page 1 of 1
The Internet of Things has certainly been the topic of much verbiage over the past year, and as the New Year opens before us, that seems destined to continue. And that is because, well, it is a huge and important topic. The general perception is that this phenomenon just sort of happened, grew out of a combination of embedded and networking technologies that reached a sort of critical mass, and now we are looking at something like 50 billion connected devices in the next few years with that number continuing to grow well beyond.
It is certainly true that the Internet of Things has come about as a result of a confluence of technological advances such as the scale of integration resulting in high performance, low power consumption, low cost, small size and integrated connectivity in very small packages. It is also generally agreed that machine-to-machine (M2M) systems—which were mostly oriented toward a specific proprietary application—gave birth to the Internet of Things, which on one level represents a broader context in which M2M can exist. But, of course, it represents much more than that as well (Figure 1).
The Internet of Things ranges from small sensors, often attached to a LAN that in turn interfaces to the Intenet and the “cloud.” The brownfield and greenfield devices represent legacy devices that have been adapted to the Internet and new devices designed specifically to connect.
Now that the Internet of Things is definitely here, we recognize its prominent features and companies are putting it to profitable use. As this happens, we can expect efforts to productize and standardize access and usage modes in ways, including products, that can help customers configure and use their interactions with the IoT. There are presently some indications that efforts in this direction are starting, which would be a signal of the maturing of this new and exciting technological phenomenon.
Wind River, a veteran company in embedded software, presently appears to be making moves in such a direction, thinking about ways the IoT can be approached and utilized to exploit its potential. Jim Douglas, senior VP of marketing for Wind River, agrees that the IoT has now matured to the point where businesses are seriously looking at their business models in light of the IoT and considering how they can move beyond simply improving productivity to actually transform their business models to take advantage of the IoT.
One of the scarier terms associated with the IoT is “Big Data.” Because huge numbers of really small things can generate really huge amounts of data, the question is how to deal with it all. One answer is that while that is true, relatively simple but very valuable things can be done with Big Data, which can be stored, analyzed in greater detail later and be applied to potentially produce even more value. According to Douglas, much can be done with data before it gets really “Big,” which means dealing with it at the aggregation nodes at the edge of the Internet/Cloud. Intelligence applied here can make profitable use of data before sending it “north” into the Cloud, and can also make needed results available to devices in its domain.
The typical scenario for many applications consists of devices and sensors (things) connected in a local area network (LAN) to one or more edge aggregation nodes. On the other side of each such node is a connection to the Internet and hence to the Cloud where the data is collected, stored and really does become big. In the Cloud, of course, it can be analyzed in a variety of ways as well.
When it comes to companies that sell and service “big items” like power plant generators, airplanes or transportation systems, Douglas points out, their actual margins are slim and they make a good part of their revenue and profit in services for things that have long amortization schedules. There are definite advantages to be gained for such companies by the access available via the Internet. These include the ability to monitor, predict and optimize performance. Much of this analysis can be done or at least pre-processed at the edge level.
For example, monitoring a fleet of aircraft could be used to statistically predict failure and address potential failures in advance by scheduling maintenance to safely keep planes in the air at a higher rate. The deep analysis or pattern discovery from data taken from a whole fleet of aircraft could be collected and analyzed in the Cloud to generate parameters that could be used to examine data coming in to the aggregation node at the edge, where it could generate alerts or examine different efficiency levels to use aircraft in situations where they could safely operate longer. Thus there is a need for sending data “north” to the Cloud as well as being able to send data back down to the edge nodes where it can be put to use and even sent all the way back to end devices if need be.
Wind River is involved in developing intelligent edge devices that can potentially range from quite simple, based on Intel’s latest low-end Quark devices, and ranging up to Atom and Core-based edge systems. With a scalable choice of edge/aggregation points in the form of ready-to-use products, customers can set up edge management layers between their intelligent connected devices and the Cloud—a level somewhat whimsically referred to as the “Fog.”
To that end, its Intelligent Device Platform (Figure 2) is a scalable, sustainable and secure development environment that simplifies the development, integration and deployment of IoT gateways. It is based on Wind River operating systems, which are standards-compliant and fully tested, as well as Wind River development tools. The platform provides device security, smart connectivity, rich network options and device management. Intelligent Device Platform includes ready-to-use components built for developing machine-to-machine (M2M) applications.
The components of Wind River’s Intelligent Device Platform intended to facilitate the implementation of scalable, intelligent edge/aggregation nodes.
Edge/aggregation points can also be a good location for Web servers that facilitate user interaction with connected devices. At the proper level of scalability, they have the resources for both security measures as well as for implementing user interfaces that can be accessed over the Internet by way of a browser. That makes it possible to put an application or proprietary server/human interface on a company’s private node with an interface that can address all the devices on that node either individually or collectively as well as their communications and data. It also provides a site for maintenance such as firmware upgrades and diagnostics that can save significantly on service visits.
Brownfield and Greenfield
Wind River also sees a huge opportunity for OEMs and developers in the giant task of bringing existing infrastructure into the world of IP connectivity. Given the previously mentioned expensive nature and long amortization schedules of much industrial equipment, it makes no sense to think about replacing it before its end of life. This is referred to as the “brownfield,” or equipment such as factory machines that were built without a thought of connecting it to IP networks and others like medical devices that were intentionally kept off the Web for security considerations.
Creating and marketing “bolt on” connectivity for a vast array of brownfield systems and devices is predictably a very attractive business proposition according to Douglas. For example, refitting rail systems involves adding equipment to monitor things as diverse as brake wear, vibration and passengers. It represents a large investment that is still considerably smaller than purchasing new rolling stock, and the potential benefits can be huge in terms of efficiency and cost savings. Newer rail cars and locomotives will undoubtedly have this intelligence and connectivity built in as a matter of course.
Such equipment and devices will be greenfield devices, which are designed from the start for both embedded control and external Ethernet connectivity. These are, of course, not just traditional equipment with added connectivity, but also innovative devices like wearable medical monitoring systems and dynamic control systems that have grown beyond traditional M2M. Many such devices will be able to request updates and parameters that have been produced by data collected, aggregated and analyzed by edge devices and made available for connected devices.
Of course, the vast volume of Big Data will continue to be streamed up to the Cloud for its own purposes. One of the easier forms of analytics will be to look for patterns in Big Data that might not have been detected before and that can be used to enhance monitoring, prediction and optimization. But intelligence at the edge looks like it can become much more significant and useful when there are ready-built systems that can be programmed and configured to take advantage of it and send it back and forth between the Cloud, “Fog” and edge as needed for all kinds of different applications.
The Internet of Things is now in place in its raw existence and is being used to multiple advantages. As more developers, businesses and people become more familiar with its capabilities and it continues to expand, we will see more innovative tools and systems to put it to use. We have only begun.