Greg’s team focuses on the digital transformation taking place throughout the industrial space. Known by terms such as Industrial Internet of Things, Industrie 4.0, Information-Driven Manufacturing, Connected Manufacturing, Smart Manufacturing, Industrial Internet, and many others, this transformation is already underway. It is powered by technologies such as machine learning, predictive analytics and Big Data, cloud computing, mobility, low-cost sensors, edge intelligence, network connectivity, and more.
Literally and figuratively the industrial world is at the “edge.” Edge computing is transforming the way data is being processed and transmitted from millions of IoT devices, globally. In industrial environments, edge computing offers the promise of getting the right device data in near real-time to drive better decisions and maybe even control industrial processes. At the ARC Industry Forum in Orlando Sastry Malladi, Co-founder and CTO, FogHorn Systems, spoke to ARC’s Greg Gorbach, Vice President, about the company’s edge offerings, how it differs from the traditional approach, and the security issues that crop up. You can watch the interview here or on YouTube.
FogHorn’s Edge Offerings
Greg began the interview by asking about FogHorn’s edge offerings, the drivers behind it, and customer benefits.
Five years ago, when FogHorn was started the term, “edge” was new to people; but now it is commonly used, said Sastry. The company entered the edge space, because industries, such as manufacturing, oil and gas, etc. “have been producing tons of data, terabytes to petabytes of data. The traditional approach of being able to process that data by transporting all of that information to a cloud environment isn’t going to cut it for many reasons: no connectivity from where these machines can send to the cloud; the amount of bandwidth that you will need to transport all of that data into a cloud environment could be cost prohibitive; and latency is involved.”
Talking about security issues, Sastry said that many industrial customers are afraid to connect their machines, heavy machinery, into any kind of internet or public cloud. Therefore, they wanted some sort of a local edge processing right where data is produced to derive actionable insight. FogHorn offers a generic edge intelligence platform that can run in a constrained, hard compute environment right at the edge; get all of the data from these different machines and sensors, process them, drive the insights so that informed business decisions can be taken.
“There are many edge solution providers, so what differentiates FogHorn from the others?” Greg asked.
Everyone’s on the bandwagon, offering what they call edge computing solutions. “But sometimes it’s misleading. Because many of them, for example, provide edge infrastructure, it could be hardware, it could be a software orchestration layer,” explained Sastry. Sometimes solution providers simply connect to the sensors, acquire and cleanse the data and send it for further processing. “That is still not true edge computing.”
Whereas, FogHorn processes the data right at the source itself. The company combines traditional digital sensors, like temperature, pressure and velocity, with some of the modern sensors (audio, video, acoustic, vibration, etc.), fuse them and then process them right at the edge in the constrained environment. The company has several patents on the technology; it invented the CEP (Complex Event Processing) engine that works on all of these different fused set of signals as well as machine learning. “Many think that machine learning can’t be done at the edge, but we’ve obviously shown to the world that it can. The combination of machine learning, AI (artificial intelligence), and the CEP engine is what we’ve invented to run in this small footprint constrained environment to process directly at the edge,” explained Sastry.
Machine Learning and Artificial Intelligence
Sometimes a plant operator may have tribal knowledge or a hunch about the reason for a potential failure. But in most cases, customers simply say that the processes are not efficient and ask the solution provider to find the root cause of these failures. “This is where we apply machine learning,” said Sastry.
Machine learning need not be done in the cloud or a big compute environment. FogHorn’s technology enables it (especially the inferences and execution part) to be done in the edge environment itself. The company supports all types of model descriptions with different standards, like PMML etc. “We edgify models developed in a cloud environment to be able to effectively run efficiently in an edge environment, and that is done through a combination of both hardware-based acceleration and a software-best acceleration,” explained Sastry.
Then comes the AI, which helps to predict a failure. Over a period of time conditions could change and issues (calibration, machine problem etc.) could arise, and the results may not be accurate. FogHorn has built modules as part of its software stack to detect this degradation in the accuracy levels, and then automatically start sending that raw data into a retraining module to automatically retrain the machine learning model.
Adoption of Edge Technologies in Digital Transformation
Earlier, customers were skeptical about adoption of edge technologies in the digital transformation journey. But after several proof-of-concepts customers have begun to realize the real value that it brings and are keen to adopt edge technologies. For this, there needs to be alignment between the business problem to be solved, the plant operators, and the budget justification. “Another challenge is to bridge the IT/OT divide,” said Sastry. FogHorn’s solutions and tooling address these issues and create an amicable working environment. Going forward, the company sees a huge adoption curve. “We are seeing a huge uptick in the adoption of edge technologies across many industries, manufacturing, oil and gas, smart buildings, transportation, including autonomous driving vehicles, energy management across the building sectors and so on. I think the potential is huge,” said Sastry.
About ARC Advisory Group (www.arcweb.com): Founded in 1986, ARC Advisory Group is a Boston based leading technology research and advisory firm for industry and infrastructure.
For further information or to provide feedback on this article, please contact RPaira@arcweb.com
About the Author: