What do these companies and organizations have in common: a supermarket chain that ensures that your favorite products are never out of stock, an oil company that prevents untold damage to the environment, and a hospital that saves lives by ensuring ambulances are never late? Simply put, they are all examples of users who leverage Internet of Things (IoT) analytics, data, and platforms.
Internet of Things (IoT) analytics is the use of data science techniques to derive actionable insights from the vast volumes of information collected by IoT devices. Common applications of IoT analytics include predictive maintenance, supply chain management, and prescriptive analytics.
As more and more ‘things’ become part of the Internet of Things, IoT analytics is developing into an integral part of the connected device universe. Data from your wearable device might be able to alert you regarding an impending heart condition. While on the factory floor, data could be used to optimize industrial processes that lead to billions of dollars in savings.
Dgtl Infra investigates the rise of IoT analytics and how it is enabling data-driven decision-making in small and large businesses alike.
Why Internet of Things (IoT) Analytics Matters
According to Gartner, 14 billion devices are connected to the Internet in 2022. These devices generate enormous amounts of data, and the biggest challenge people, businesses, and organizations face is to sustainably leverage the data created by these networks of interconnected systems.
Presently, a surge in the amount of data produced by IoT devices is occurring, with volumes expected to reach 79.4 zettabytes by 2025. This represents a 484% increase from the total data output from IoT connections in 2019. To put this figure into perspective, one zettabyte is one billion terabytes or one trillion gigabytes.
IoT analytics has the ability to process vast volumes of data to detect patterns and make predictions, effectively converting raw data into useful intelligence and insight.
No surprises then, that the IoT analytics market is valued at $28.9 billion in 2022, with forecasts expecting the market to reach $81.7 billion by 2026, representing a nearly 30% compound annual growth rate (CAGR).
IoT devices capture data in a variety of formats, such as structured, semi-structured, and unstructured data. Interestingly though, most of this information is collected as unstructured data. This is one of the primary differences when compared to traditional analytics, which largely deals with structured data.
As sensors may be affected by physical processes, they are likely to have missing data points, corrupted messages, and incorrect readings – unlike data entered by humans.
As shown below, due to a number of factors, IoT analytics often involves more data, more complexity, and more automation.
What Makes IoT Analytics Different?
Data can be collected in the form of analog signals, sensor readings, device health metadata, or large files consisting of images and videos. Since there is no homogeneity in the data collected from different IoT devices, there is no ‘one-size-fits-all’ approach to storing and analyzing IoT data.
Some companies might need to use the data only at the time of collection and have little use for it afterward. For example, applications that map weather in real-time fall under this data usage pattern.
Otherwise, companies will store their collected data for analysis at a later point in time. Everything from traffic patterns at busy intersections, to retailers predicting demand during a busy holiday season, can be categorized under this data usage pattern.
Big Data and Internet of Things (IoT) Analytics
Big data is allowing us to make sense of billions of real-time, unstructured data points recorded by IoT devices. These technologies and methods for massive data analysis turn data sets that are too large or complex to be processed by commonly-used software tools, into information that guides businesses on how to optimize their processes.
As an example, big data explains why connected cars, with more than 40 microprocessors and dozens of sensors, will send 25 gigabytes of data to the cloud every hour. Car makers are able to capture and analyze data on everything – speed, route, wear & tear, and even road conditions.
Types of IoT Analytics
Overall, there are four main types of Internet of Things (IoT) analytics:
- Descriptive Analytics: aggregates and makes sense of real-time data coming from IoT devices
- Diagnostic Analytics: builds on the initial understanding from descriptive analytics and explains why things are happening
- Predictive Analytics: incorporates machine learning (ML) capabilities to predict future events based on past data
- Prescriptive Analytics: provides additional insights and actions that can be taken to improve the efficacy of descriptive, diagnostic, or predictive analytics
Examples of IoT Analytics
The greatest potential for IoT analytics lies in the Industrial Internet of Things (IIoT). Organizations are able to deploy sensors in everything from manufacturing equipment, to pipelines, to weather stations, and delivery trucks – for the purposes of collecting and analyzing data.
Audi – Manufacturing and IoT Analytics
Audi’s Neckarsulm factory is an illustrative example of the power of IoT analytics in manufacturing. Here, up to 1,000 cars are assembled in the plant every day, with 900 robots tasked with performing 5,000 welds in every car. Indeed, that is 5 million welds in a single day of production, making it impossible to perform manual quality control operations.
In-partnership with Intel, Audi was able to create a predictive analytics solution that transformed factory data from welding gun controllers into valuable insights. The data-driven weld inspection process resulted in improved accuracy and a 30% to 50% reduction in labor costs.
Levi Strauss – Retail and IoT Analytics
Inventory distortion is a problem in the retail industry, which is usually evident in the form of overstocking, stock-outs, and shrinkage. These situations result in over $1 trillion worth of losses every year for retailers worldwide.
Levi Strauss partnered with Intel to deploy a real-time inventory monitoring solution in their flagship store in San Francisco, California. By using radio-frequency identification (RFID) tags on all items, Levi’s was able to locate and account for every item on the sales floor at any point in time.
Data gathered by the RFID tags was then sent to the cloud for analytics purposes, leading to 100%-visibility on what inventory was on the shelf and improved the company’s understanding of customer buying behavior.
BP – Oil & Gas and IoT Analytics
BP, a British multinational oil & gas company, decided to equip 650 out of its thousands of oil wells with General Electric (GE) sensors. Specifically, each well was outfitted with 20 to 30 sensors to measure pressure, temperature, and other crucial metrics – sending about 500,000 data points to the cloud every 15 seconds. In turn, this solution allowed BP to predict oil well flows and the life of each well, which ultimately, helped the company determine the performance of an entire oil field.
Deep Sky – Agriculture and IoT Analytics
Agriculture and IoT analytics implementations have been showcased through a partnership between Deep Sky, a vineyard for wine, and Google Cloud. Using Google Cloud-powered hardware throughout their vineyard in Arizona, Deep Sky was able to monitor metrics such as water flow and soil moisture to improve grape yields.
Overall, this IoT analytics solution increased the vineyard’s energy savings by 15%, labor efficiency by 30%, and crop efficiency by 50%, while reducing costs associated with human error by 75%.
Edge Analytics in IoT
Usually, IoT analytics is not applied on raw data captured at the device-level. Incorrect readings and duplicate values are quite common in sensor-generated data and they are filtered-out at the point of data collection. Gateway devices or the IoT devices themselves often determine which data needs to be transmitted upstream to an analytics engine.
Edge analytics is the process of collecting and analyzing data in real-time, directly from the IoT devices generating the information. In so doing, actionable insights can be made as close as possible to the devices producing the data.
One of the main factors driving the use of edge analytics is privacy, especially when IoT devices are capturing sensitive information like GPS data or live video streams.
Additionally, edge analytics offers low-latency and reduces the bandwidth requirements of a network, since the same amount of data does not have to be transmitted between a device and the cloud.
READ MORE: What is an Edge Data Center? (With Examples)
Healthcare – Edge Analytics
Healthcare is one of the largest beneficiaries of edge analytics. Given the ongoing global shortage in health workers, technology is stepping-in to fill the void with edge computing and analytics.
For example, portable magnetic resonance imaging (MRI) machines can now capture brain scans at a patient’s bedside, while machine learning (ML) algorithms are processing the scans and identifying anomalies, allowing radiologists to generate reports in real-time.
Video – Edge Analytics
Sony’s REA-C1000 Edge Analytics Appliance is an example of a video analytics IoT device designed for consumers and businesses. Specifically, this device incorporates artificial intelligence (AI) into video processing technology to deliver more impactful content while a user is giving a video presentation.
In terms of functionality, the device has a handwriting extraction feature powered by augmented reality (AR) technology to ensure that any words or diagrams virtually written on-screen remain in full view for the audience. Additionally, the appliance has a chroma key-less CG overlay feature, which creates a green screen so that a static, virtual background can replace the presenter’s real-life background. Finally, the device’s capabilities also include PTZ auto tracking and close-up by gesture.
IoT Analytics Platforms
IoT analytics platforms make it possible for businesses to aggregate, analyze, and visualize the massive streams of data captured by IoT devices. Particularly, Google Cloud, Microsoft Azure, Cisco, ThingSpeak, and Amazon Web Services (AWS) are some of the leading providers of IoT analytics platforms.
Google Cloud IoT Core
Google Cloud IoT Core is a fully-managed cloud service for connecting, managing, and ingesting data from thousands of Internet of Things (IoT) devices spread out around the world.
Using Google Cloud’s compute, storage, and networking infrastructure as the backbone, the IoT Core service is able to offer edge messaging services, device management, analytics, and the ability to derive insights, using Google Cloud’s artificial intelligence (AI) and machine learning (ML) tools.
Azure Stream Analytics
Microsoft delivers a serverless, real-time analytics engine with its Azure Stream Analytics solution. This service facilitates analytics to be run on streaming data, from the cloud to the edge.
Azure Stream Analytics is designed for mission-critical workloads which can be run with sub-1 second latencies. Furthermore, the service is extendable with custom code and machine learning (ML) capabilities for more advanced scenarios, such as anomaly detection.
Cisco Kinetic facilitates the connection of distributed IoT devices to the network. The service then enables data to be extracted, normalized, and securely moved from those devices to distributed applications. Additionally, this IoT analytics platform can enforce governance policies defined by data owners, so they can control which data goes where, and when.
Overall, Cisco Kinetic’s integrated IoT analytics platform allow users to i) deploy and manage Cisco gateways from a remote location and access devices over a secure VPN connection, ii) transform and filter sensor data, while sending the results to the cloud, and iii) access compute on distributed nodes of the network – from edge to destination.
ThingSpeak is an open-source IoT analytics platform that allows for rapid prototyping and deployment of an IoT system with data analytics. The service allows users to aggregate, visualize, and analyze live data streams in the cloud by sending data to ThingSpeak.
Features of ThingSpeak include RESTful and MQTT APIs for transmitting IoT device data, as well as MATLAB widgets for analysis and visualization. Also, ThingSpeak offers compatibility with different IoT platforms and devices, such as NodeMCU, Raspberry Pi, and Arduino.
Amazon Web Services (AWS) IoT Analytics
Amazon Web Services (AWS) IoT Analytics is a fully-managed service that enables complex analytics to be run on large volumes of IoT data.
AWS IoT Analytics automates all the challenging steps involved in analyzing IoT device data. As shown below, the service filters, transforms, and enriches IoT data before storing it in a time-series data store for analysis.
Subsequently, users can analyze their data by running ad hoc or scheduled queries using a built-in SQL query engine, as well as perform more complex analytics. Finally, AWS IoT Analytics enables businesses to run machine learning (ML) analysis, using its pre-built models for common IoT use cases.
How AWS IoT Analytics Works
AWS IoT Analytics is a pay-as-you go service that scales automatically to support up to petabytes of Internet of Things (IoT) data. With this service, users can analyze their entire fleet of connected devices without managing hardware or infrastructure.
Use Cases for AWS IoT Analytics
Below are four examples of how Internet of Things (IoT) data can be harnessed using the AWS IoT Analytics platform.
AWS IoT Analytics provides templates for users to build predictive maintenance models, and apply them to their fleet of IoT devices. For example, AWS IoT Analytics can be used to predict when heating and cooling systems are likely to fail on connected cargo vans, so that the vehicles can be rerouted to prevent shipment damage.
Proactive Replenishing of Supplies
AWS IoT Analytics allows users to build IoT applications for real-time monitoring of inventories. For example, a food and drink company can use IoT analytics data from vending machines and proactively reorder merchandise whenever stocks are running low.
Process Efficiency Scoring
Building applications to constantly monitor and improve the efficiency of different processes can be done through AWS IoT Analytics. For example, a mining company can increase the efficiency of its ore trucks (e.g., used to mine copper or iron), through IoT device monitoring / analytics, and, in turn maximize the load for each trip.
AWS IoT Analytics can enrich IoT device data with contextual metadata using AWS IoT registry data or public data sources. In this way, data analysis can factor in time, location, temperature, altitude, and other environmental conditions.
As an example, to determine when to water the fields, irrigation systems might enrich humidity sensor data with data on rainfall, further improving the efficiency of water usage.