FogHorn blows its trumpet at the edge

With $47.5mn raised since its foundation in 2014, FogHorn is one of the better endowed start-ups specializing in the fast-growing field of industrial edge computing. It takes its name from Fog Computing, a term coined a little earlier by Cisco to describe processing in a layer between the enterprise or consumer and the cloud.

This area attracted a new breed of start-up, but mostly still with an emphasis on bringing data into the cloud where possible so that enough affordable resources could be brought to bear upon it. This led FogHorn’s co-founder and CEO David King to decry such technologies as being little more than glorified store-and-forward mechanisms.

Although some do more than that, this deficiency in many approaches has allowed FogHorn to set out its stall as the champion of intensive low footprint edge computing, capable of delivering actionable analytics in real time if required. This is clearly not possible if data must be brought back to a centralized cloud-based system a long way from the industrial source of the data.

Edge computing in some cases is deployed to cut latency or for security reasons to prevent sensitive data leaking back into the cloud. FogHorn’s focus though is on the need for rapid analytics-based decision making, which it said has potential across a wide range of industrial processes from oil rigs to automotive manufacturing. It addresses the lack of resources often available, and in some cases intermittent and low bandwidth connectivity.

Its two main contributions are a stripped down miniaturized complex event processing (CEP) engine and ML-based platform for deriving real time insights at the edge. It has also developed a domain specific language (DSL) to define failure conditions and detect complex events of interest from the multitude of incoming sensor streams of data. Data pre-processing is crucial for many of the use cases to reduce data size and prepare it into a form fit for ML based analytics tools to execute.

The company in July 2017 released the final version of its edge computing engine called Lightning Edge ML, which was designed to exploit existing models and algorithms as well as run new ones developed by FogHorn. The firm strived to make the ML accessible to non-technical staff, with tools to enable models to operate on live data streams produced by industrial control systems without intimate knowledge of the system.

This product incorporated the fully stripped-down version of the software which was now capable of running in highly constrained compute devices such as PLCs (programmable logic controllers), Raspberry Pi single-board computers, and ruggedized IIoT (Industrial IoT) gateways.

FogHorn is up against some powerful competitors, such as Amazon’s Greengrass – software for edge gateways and appliances that was launched in June 2017. Another with wider applicability is Microsoft’s Azure Stream Analytics, an event-processing engine designed to process and analyze high volumes of streaming data from devices, sensors, web sites, social media feeds and other applications, incorporating algorithms for pattern identification.

However, FogHorn aims to score by focusing on specific use cases that span various sectors, especially predictive maintenance.