In light of the global pandemic, data accessibility, visibility and interconnectivity have become critical components of business agility strategies implemented during a turbulent time. Indeed, IoT adoption has surged over the past few years and is continuing its steady upward path. Unfortunately, as organizations scale-out existing IoT efforts during a period with more budget limitations and less room for trial-and-error, many are running into a critical challenge.
Organizations are seldom prepared for the tsunami of data involved in full-scale IoT and IIoT deployments. Over the next few years, analysts estimate there will be 41.6 billion connected IoT devices generating 79.4 zettabytes of data. What’s more, about 25% of all data will be real-time in nature, adding complexity to the list of challenges organization must plan for and overcome. This article will highlight critical gaps within current IoT projects, why these gaps matter and how edge computing capabilities will increase IoT scalability and success moving forward.
State of cloud
The majority of modern organizations rely on a mix of cloud and legacy platforms to meet infrastructure needs. However, analyzing raw data from IoT sensors in the cloud is often expensive and time consuming due to data transport and processing costs. Cloud latency, bandwidth and security challenges continue to be a significant roadblock, specifically for industrial industries that produce high-fidelity, raw machine and IoT sensor data. As a result, organizations frequently resort to using down-sampled or time deferred data to balance cost and timeliness, making it easy to miss anomalies in data.
Although the cloud is an effective data modeling and learning portal, it lacks real-time capabilities needed for mission critical IoT applications in markets such as manufacturing, oil and gas, and transportation due to transmission and ecosystem considerations.
Introducing edge-first deployments
By implementing edge-native solutions, organizations can ingest, enrich and analyze data locally, execute machine learning models on cleaned data sets and deliver enhanced predictive capabilities. Edge computing is critical for a wide variety of IoT-powered use cases where real-time capabilities are required. Think worker health and safety monitoring, including temperature, face protection and social distancing. Industries with security concerns or limited access to bandwidth, such as mining and fleet, are also greatly benefiting from edge computing.
Keep in mind, edge-first IoT initiatives do not eliminate all cloud involvement. In fact, edge solutions rely on the limitless resources of cloud environments to train and improve existing machine learning models. Edge devices that execute machine learning on live streaming data must regularly check for model accuracy and environmental changes over time. As model accuracy drifts, insights are then sent back to the cloud including data that represents unusual activity warranting retraining of the current models. Once the models are fine-tuned they are pushed back to the edge, which results in a constant, closed-loop process that generates much higher quality predictive insights to improve asset performance, process improvements and product quality.
By running cloud-edge versions of machine learning models in real-time, organizations enable the ability to act, react and pro-act to events of interest at the source. This ensures a harmonious interplay of IoT, edge and cloud, using the strengths of each ecosystem. Additionally, cloud-edge hybrid solutions prevent cloud lock-in because different use cases can publish insights into one or more public and private clouds.
Cloud-edge benefits for IoT
Cloud-edge hybrid initiatives transform real-time IoT data into actionable insights related to production efficiency and quality metrics that can be used by operations managers to reduce unplanned downtime, maximize yield and increase machine utilization. For example, using edge-cloud hybrid strategies, factories can improve product quality. By analyzing IoT sensor data in real time, organizations can identify any values falling outside of previously defined thresholds and rules, build and train a machine learning model to identify root problem causes, and deploy the machine learning model to automatically stop the production of defective parts.
Additionally, edge-cloud insights allow smart building operations personnel to monitor energy use and proactively modify operations to avoid outages due to overworked energy systems. Rather than relying on delayed insights from cloud-only systems, managers and operators can access insights in real time to more quickly identify the root cause of IoT-powered building system disputes and ultimately reduce overall downtime.
All IoT Agenda network contributors are responsible for the content and accuracy of their posts. Opinions are of the writers and do not necessarily convey the thoughts of IoT Agenda.