A couple of weeks ago, I spent a few hours at GE Digital’s headquarters in San Ramon, CA. It was a great overview by several executives of how GE is using their Predix platform to create software to design, build, operate, and manage the entire asset lifecycle for the Industrial IoT. A big part of this transformation for GE involves hiring tons of software developers, acquisitions, and partnerships.
One of those partnerships is with Silicon Valley based FogHorn Systems (GE Ventures, Dell Ventures, March Capital and a few others are investors). FogHorn is a developer of “edge intelligence” software for industrial and commercial IoT applications. FogHorn and GE are working very closely on many IIoT customer use cases, across verticals, bolstered by the integration of FogHorn with Predix.
I turned to FogHorn Systems CEO David C. King to learn more about edge intelligence software for the Industrial IoT. David has been at the helm of FogHorn since 2015, a year after its founding. Prior to FogHorn, David co-founded AirTight Networks, Inc., a technology leader in secure cloud-managed Wi-Fi. Before AirTight, he served as Chairman, President and Chief Executive Officer of Proxim Inc., a pioneer in WLANs and the first publicly traded Wi-Fi company, from 1993-2002.
Lots of talk about the edge in IoT. It’s my smartphone and my doorbell, as well as the sensor on a traffic light or a wind turbine. What exactly is the edge of the network and how do you define it?
We define edge as the closest compute point that can process real time streaming data. So in your case, all three -- phone, doorbell, sensors -- are edges because you can bring compute to the data on any of these platforms. The question is what compute is possible? The single variable filtering that you can do on a sensor is very simple when compared to the complex Machine Learning models that can execute on your phone.
Analytics is done in the data center or cloud. You claim to do this at the edge now. Please describe your offering.
FogHorn has developed a tiny footprint complex event processor (CEP) that provides advanced streaming analytics, and machine learning capabilities at the edge. This powerful combination of being able to pre-process, cleanse the data and execute ML models, all in real-time, brings the power of big data analytics to the edge. The FogHorn software platform is highly flexible and can be easily scaled to optimize for footprint and/or feature needs.
Tell us about a customer you’re working with and how they are applying your technology.
FogHorn Lightning is an extensible platform currently used by customers from Manufacturing, Oil & Gas, Power & Water, Renewable Energy, Mining, Transportation, Smart Buildings/Cities and other industrial verticals. The deployment patterns range across gateways, PLCs, to ruggedized servers in production, at Fortune 100 sites. A common implementation of FogHorn Lightning is product quality inspection, predictive maintenance, real time health monitoring. Customers are seeing immediate business value; e.g. identifying defects in the early stages of manufacturing reduces, scrap and increases yield. Additionally, there is a trend to FogHorn to generate new streams of revenue by providing real-time smart maintenance for their end customers.
When compared to software-defined IIoT smart gateways, there are still millions more hardware-defined M2M gateways out there. At what point do we cross the chasm to smarter gateways, and where are we now in this cycle?
We are still very early in adoption of IIoT technologies. Understandably, typical industrial sectors are conservative, and have much longer adoption curves. However, we are beginning to observe that it the ROI from edge intelligence is accelerating customer demand for FogHorn. We will cross the chasm once industries identify key use cases that generate new revenue streams, which is still about 3-5 years away.
You can’t talk about IoT without talking about security, and it’s even more important in the industrial sector. How do you address security concerns for your customers and what does the industry need to do to make IoT more secure?
Yes, you are right. When you think of IoT, especially IIoT, security is a top concern. Hacks such as “Devil’s Ivy” will become everyday events with increasingly connected devices. At FogHorn, our edge intelligence software runs very close to the data source, and is local to the asset. This implies that we are secure (like the assets) behind firewalls, and in a DMZ layer. And because most of our processing is done locally, we are less vulnerable to malicious hacks that occur when connected.
Because IIoT is still such a nascent set of technologies, we caution users to deploy solutions after thoroughly weighing the business value, and convenience versus security risk factors. My guiding question before any deployment: “Can I do this locally, without connecting to an external network?”. The answer is usually yes, and if otherwise, you should probably talk to us.
How can companies make their industrial processes better?
We understand that today’s industrial processes are highly complex and advanced, with many moving parts. While it may seem humanly impossible to optimize it any more without help from technology, we believe that a key asset is still untapped: your operator! Companies will start seeing incredible improvements once they translate the tribal knowledge on the plant floor into actionable insights. This can be further supplemented by techniques from machine learning, and artificial intelligence, to tease out the known unknowns, and also, the unknown unknowns.
Anything else you’d like to add?
FogHorn is redefining edge intelligence for IIoT. A year ago, we started our journey as a company that did analytics on tiny footprint devices. Today, we have accelerated the transition to Machine Learning at the edge, and are very are excited about the market validation. With our Operational Technology focus, we are looking forward to defining new business models, and delivering transformational value for our industrial customers.