Section: New Results
AI across the digital continuum
Machine Learning in the context of Edge stream processing.
Participants : Pedro de Souza Bento Da Silva, Alexandru Costan, Gabriel Antoniu.
Our research aims to improve the accuracy of Earthquake Early Warning (EEW) systems by means of machine learning. EEW systems are designed to detect and characterize medium and large earthquakes before their damaging effects reach a certain location.
Traditional EEW methods based on seismometers fail to accurately identify large earthquakes due to their sensitivity to the ground motion velocity. The recently introduced high-precision GPS stations, on the other hand, are ineffective to identify medium earthquakes due to its propensity to produce noisy data. In addition, GPS stations and seismometers may be deployed in large numbers across different locations and may produce a significant volume of data consequently, affecting the response time and the robustness of EEW systems.
In practice, EEW can be seen as a typical classification problem in the machine learning field: multi-sensor data are given in input, and earthquake severity is the classification result. We introduce the Distributed Multi-Sensor Earthquake Early Warning (DMSEEW) system, a novel machine learning-based approach that combines data from both types of sensors (GPS stations and seismometers) to detect medium and large earthquakes.
DMSEEW is based on a new stacking ensemble method which has been evaluated on a real-world dataset validated with geoscientists. The system builds on a geographically distributed infrastructure (deployable on clouds and edge systems), ensuring an efficient computation in terms of response time and robustness to partial infrastructure failures. Our experiments show that DMSEEW is more accurate than the traditional seismometer-only approach and the combined-sensors (GPS and seismometers) approach that adopts the rule of relative strength.
These results have been accepted for publication at AAAI, a "A*" conference in the area of Artificial Intelligence [21].
ZettaFlow: Unified Fast Data Storage and Analytics Platform for IoT
Participants : Ovidiu-Cristian Marcu, Alexandru Costan, Gabriel Antoniu.
The ZettaFlow platform (system of systems) provides a high-performance multi-model analytics-oriented storage and processing system, while supporting publish-subscribe streams and streaming, key-value and in-memory columnar APIs [16].
The ZettaFlow project is funded by EIT Digital from October 2019 to December 2020. It includes three partners: Inria for the platform development, TU Berlin for edge to cloud IoT optimizations with microservices, and Systematic Paris Region for the go-to-market strategy.
Our goal is to create a startup that will commercialize the ZettaFlow platform: a dynamic, unified and auto-balanced real-time storage and analytics industrial IoT platform. ZettaFlow will provide real-time visibility into machines, assets and factory operations and will automate data driven decisions for high-performance industrial processes.
ZettaFlow will bring a threefold impact to the IoT market.
-
Enable novel real-time edge applications that truly automate manufacturing, transportation and utilities processes.
-
Reduce deployment efforts and time-to-decision of IoT edge-cloud applications by 75% through automation, unified dynamic data management and streaming analytics.
-
Reduce human costs for monitoring and engineering (through edge intelligence) and IoT hardware costs by 50% through unified data collection/storage/analytics.