Work package 1

Big data management

M6-M48

Objectives

Provide a set of innovative solutions for the management of IoRT data (heterogeneous, complex and voluminous) over the edge-fog-cloud architectures. For data collection: provide (i) a set of configurations for IoRT devices and networks that grant for a reliable sensing and transmission process, (ii) Methods to improve the quality (precision, uncertainty, missing values, outliers, etc.) for a better and effective decision-making process (WP2-WP3) and (iii) New storage and querying technologies allowing an effective exploitation for analytics methods (WP2-WP3).

Task 1.1 Data sensing and processing at the edge

Robots and sensors have limited storage and computation capabilities, and they can run in rural zones covered by communication networks of low throughput. This task will define new methods to mitigate problems with network coverage and throughput and minimize data transmission costs, run lightweight computation algorithms/models in order to grant data reliability, and usage of data standards to support interoperability.

Task 1.2 Data quality management.

This task proposes a set of methods to detect, assess and/or improve data reliability for complex data (e.g. 3D, hyperspectral images, acoustic signals, GPS,etc) acquired by fixed sensors and robots’ embedded. They will be based on AI and other computation methods (e.g., descriptive statistical methods), integrating the field expertise of end-users (such as reinforcement learning). New data quality methods will be developed at the integration stage by means of AI and statistical methods, by means of two architectural integration approaches Extract-Transform-Load (ETL) and Extract-Load-Transform (ELT) in order to best fit with used edge-fog-cloud architectures.

Task 1.3 Data storage and querying

New logical and physical schemas for the storage and querying will be studied: multi-model, relational and NoSQL polyglot, and spatio-temporal database systems.

Task 1.4 Data management architecture

A reference edge-fog-cloud IoRT data management architecture will be developed.

Task 1.5 Experiments

Proposals will be put to the test against real data and with real end-users.