X

The browser version you are using is not recommended for this site.
Please consider upgrading to the latest version of your browser by clicking one of the following links.

FogHorn Lightning - AI at the Edge

FogHorn

Overview of FogHorn Lightning

FogHorn’s Lightning™ Edge AI platform embeds artificial intelligence locally, at or near the source of streaming sensor data. The highly compact and feature-rich Edge AI solution delivers unprecedented low latency for onsite data processing, real-time analytics, ML and AI capabilities in the smallest compute footprint. FogHorn offers a fully-integrated, closed loop edge to cloud solution, rapidly iterating ML models to adjust to changing operating conditions.

FogHorn

VEL (CEP)

FogHorn’s VEL Complex Event Processor (CEP) was engineered specifically for industrial edges:

VEL (CEP)

Performs real-time analysis of disparate streams of sensor data
Optimizes constrained and diverse compute environments that have limited or no connectivity
Simplifies interoperability with existing OT systems
Simultaneously performs complex pattern recognition on high frequency and inherently asynchronous streaming data
Detects events in real-time, enabling immediate closed loop control actions, resulting in significant cost savings
Handles machine learning pre and post processing (cleansing, filtering, normalizing, and contextualizing streaming sensor data) optimizing it for machine learning performance

EdgeML

The Lightning Edge AI platform's Edgification™ technology morphs machine learning models to execute efficiently at the industrial edge:

EdgeML

The VEL CEP engine, in addition to pattern detection and analytics, provides data pre and post processing, preparing streaming sensor data for machine learning
Reduces machine learning model size and required memory by more than 80%. This enables fast and efficient ML execution in resource-constrained environments
Import existing models (PMML, Python) seamlessly into the platform, bridging OT-IT divide
The cloud plays a critical role in ML model creation and training, especially for deep learning models. Once trained, the model can be edgified and pushed to the edge. EdgeML enables an iterative closed-loop edge to cloud machine learning cycle
Execute neural net models (example: video analytics) in constrained compute environments with the same powerful results as using the cloud

OT-Centric Toolkit

The Lightning platform offers a robust set of tools to help clients easily create new analytic expressions with a drag and drop UI, simulate sensor traffic, visualize streaming sensor data, and centrally manage FogHorn deployments in a scalable fashion. These include:

OT-Centric Toolkit

FogHorn Manager – A browser-based tool that lets deploy configuration or custom applications across thousands of Edges, create analytic expressions, and create data publications to send the insights to a preferred cloud storage location
REX – A multi-purpose tool that simulates sensor traffic to debug production issues in industrial IoT environments
VIZ – Lets you visualize real-time streams. It can be used to validate sensors, troubleshoot problem input sources, visualize the results of analytic expressions, or view the output of machine learning algorithms
VEL Studio – Gives users the ability to author and debug analytic expressions using a drag and drop interface. Additionally, it provides templates that are commonly used in industrial settings to accelerate the creation of new analytic expressions. The visual debugger helps validate and troubleshoot expressions using simulated and authentic production data

Prerequisites

This document assumes that Intel SmartEdge-Open(formerly known as OpenNESS) was installed with Minimal Flavor. FogHorn requires that the Longhorn storage class ( http://www.longhorn.io/ ) is installed on the cluster. Therefore, please run as a prerequisite:

$ helm install longhorn longhorn/longhorn --namespace longhorn-system

Usage

Repository

The required FogHorn Lightning edge helm charts are available and can be requested by dropping a mail to learning@foghorn.io. You will receive two helm charts (foghorn-registry-setup and foghorn-edge) together with the license for installing the software.

Install the FogHorn helm charts

Installation requires to install the foghorn-registry-setup helm chart before the foghorn-edge helm chart as shown below:

$ helm install foghorn-registry-setup . --namespace foghorn
$ helm install foghorn-edge . --namespace foghorn

Install FogHorn Manager

A running instance of FogHorn Manager is required. This can run in the cloud or on the OpenNESS cluster. To install FogHorn Manager on the OpenNESS cluster please use fh_util.sh which you will receive together with your license:

$ fh_util.sh install kubernetes_fhm

Uninstall the FogHorn Edge Chart

$ helm uninstall foghorn-edge . --namespace foghorn
$ helm uninstall foghorn-registry-setup . --namespace foghorn

Additional Information

FogHorn is the first edge-native AI solution in the market, delivering transformational business results.

Actionable insights in real-time for secure, on-site intelligence

The Lightning™ Edge AI Platform brings the power of AI to the edge, securely inside your business, where data is created and decisions are made. It delivers real-time analytics on high volumes, varieties and velocities of live sensor and machine data, and is optimized for limited compute and connectivity.

Higher quality predictive insights to drive organization-wide improvements

FogHorn’s Edgification™ technology allows machine learning (ML) models to be edgified (optimizing models for hyper-efficient edge execution), reducing model size by up to 80%. This moves ML and AI as close to the data source as possible, utilizing all of the available raw data, and enables faster and more efficient operation with better insights.

Reduce comms, cloud processing and storage costs by 100-1000x

Processing the majority of industrial sensor data at the edge radically reduces data transport and storage costs. FogHorn’s subscription, not consumption-based pricing further boosts savings. Additionally, less networking and security resources are needed. This approach is especially powerful in data intensive use cases, and ideal to handle the growing use of video sensors.

Works with all major cloud providers

FogHorn’s Lightning is cloud agnostic, facilitating hybrid and multi-cloud deployments. This avoids cloud “lock in”, increasing bargaining power and reducing sourcing costs. Many use case specific factors together can make one cloud solution much more cost effective than others, so keeping your options open is critical for current and future needs.

Taps easily into OT tribal knowledge

Foghorn Lightning is essentially a non-intrusive operational analytics and intelligence layer that runs on top of existing control systems. It provides an OT-centric toolkit for authoring and expressing events of interest (related to condition monitoring, predictive maintenance, failure detection, etc.). The tools easily enable expressing operator’s tribal knowledge into deriving actionable insights.

Leverages existing small footprint and controller hardware

Designed to run on existing industrial control systems and highly constrained edge compute devices, FogHorn enables real-time analytics and ML as close to the data source as possible. This minimizes investments in heavy compute or new industrial control systems, and the software can be right-sized based on available compute, from MBs of memory to multi-core servers.

Optimized for Intel® Smart Edge

Demo or Trial

Contact learning@foghorn.io to request a demo or trial

Where to Purchase

FogHorn Lightning™ Edge AI platform

For more information about FogHorn Lightning™ visit www.foghorn.com or email learning@foghorn.io

Related Material

-
-