Big Data in IoT
Download as PPTX, PDF0 likes384 views
The document discusses the generation and management of big data in IoT, highlighting the immense volume of signals produced by industrial plants and enterprises. It covers various analytics techniques, including machine learning and process optimization, and compares Industry 3.0 and 4.0 in terms of data approach. Additionally, several case studies are presented, illustrating practical applications of big data in sectors like energy, nuclear monitoring, telecommunications, and e-government services.
Big Data in IoT
- 1. Big Data Cases in IoT Victor Polyakov, CEO of Tibbo Systems
- 2. Things Produce Really "Big" Data • IoT devices are the most dynamic data generators • An industrial plant typically generates 2,000-10,000 signals per second, while an enterprise - up to 100,000 signals per second, that is, up to 10 billion per day • Compare it with the number of bank transactions or applications for credit score • Primary problem is storage • Data lake based on the IoT platform is a combination of relational, NoSQL, key-value, graph, round-robin, file and other DBMS types - 2 -
- 3. IoT Analytics Analytical modules that "fish" patterns and reasons for management decisions in the data lake: • Machine learning (prediction, anomaly detection, classification) • Stream data processing (StreamSQL), correlation and event root cause search • Performing what-if analysis • Automated work with multidimensional cubes (OLAP) • Designing dynamic process and object models - digital twins - 2 -
- 4. How Is Industry 3.0 Different from Industry 4.0? • Obvious first answer (decision making without human involvement) is not correct • Different approach to data • Digital enterprise systems rarely interact directly, instead they analyze the experience of previous years and optimize their behavior based on the data lake • Process bus is a river along which data flows into the lake • At the moment, almost any enterprise can get economic benefits from in-depth analytics • IoT merges with Big Data and analytics - 2 -
- 5. Big Data Cases in IoT • Process bus and smart metering for Transneft Energy: 40 servers on five levels, about 2 million channels, hundreds of thousands of signals per second • Nuclear research reactor monitoring, Saint-Petersburg Institute of Nuclear Physics: 18K tags, 10Hz polling rate, data collection from National Instruments real-time chassis, lab analysis of source data (oscillograms) without aggregation • Sugar beet storage conditions monitoring, Rusagro: predictive models based on data collected from thousands of sensors, weather forecast and several other metrics - 4 -
- 6. More Big Data Cases in IoT • Monitoring of a large telecom network: 500,000 units of network equipment generating about 100 million events a day, with about 10,000 significant events remaining after automatic correlation • Integrated automation system of the Huelva Port Authority in Spain: around 2 million events about loading, unloading, changing storage areas and internal state of containers per day received from RFID readers, complex analytics for optimization of logistics processes • Kazakhstan e-government services monitoring: each application in each part of the IT infrastructure is a separate object for analysis - 4 -
- 7. Thanks for your attention! Victor Polyakov, victor@tibbo.com