Introducing the EMERALDS Toolset
Discover EMERALDS advanced urban mobility data analytics. Our architecture's core principles ensure security, performance, and seamless experiences. The first version of the EMERALDS Toolset will be developed by June 2024. The various containerised services will be tested by the EMERALDS Use Cases and Early Adoption Demonstrators with an updated Toolset available by October 2025.
Extreme-scale Mobility Data Analytics (MDA) at the CC
Extreme Scale Mobility Data Analytics (MDA) at the Compute Continuum involves advanced techniques for analysing large and diverse mobility data sets. It tackles challenges like bursty data generation and skewed distributions by rethinking traditional analytics paradigms. Innovative approaches such as in-situ data processing and distributed computing are employed to efficiently analyse mobility data across the compute continuum, from edge devices to the cloud.
Privacy-aware in situ Data Harvesting
"Privacy-aware in situ data harvesting" refers to EMERALDS' innovative approach to collecting data at the edge while being conscious of privacy concerns. It involves merging data streams, allowing user customization, and implementing real-time privacy checks to create a dynamic and privacy-conscious data processing framework.
Extreme-scale Cloud/Fog Data Processing
Extreme-scale Cloud/Fog Data Processing refers to the capability of processing exceptionally large and complex datasets in cloud and fog computing environments, with a particular emphasis on scalability, efficiency, and real-time or near-real-time processing. The EMERALDS services will be tailored for urban mobility data, to support ultra-scalable query processing algorithms over road networks, from similarity search of spatio-temporal data to more complex processing tasks for trajectory joins, discovery of hot-spots, as well as for supporting mobility analytics.
Mobility Data Fusion and Management
Mobility Data Fusion and Management is a multidisciplinary approach that involves the collection, integration, processing, and analysis of diverse data sources related to mobility, such as location data from connected vehicles, GPS-enabled devices, and other sensors. It aims to efficiently manage and make sense of large volumes of mobility data to derive valuable insights, support decision-making, and improve transportation and urban planning. In particular, EMERALDS project will investigate trajectory simplification/smoothing methods to reduce the data size, lossless data compression, depending on the sampling rate and the change in temporal properties.
Active & Federated Learning over Mobility Data
Active & Federated Learning & Explainable AI (XAI) represent a trio of advanced techniques in urban mobility artificial intelligence and machine learning, with focus on enhancing the efficiency of data labeling by selectively choosing data points that are most informative for improving a machine learning mode (Active Learning), addresses privacy concerns by allowing machine learning models to be trained collaboratively across decentralized edge devices or servers, preserving data privacy while leveraging diverse data sources in urban mobility applications (Federated Learning) and making machine learning models transparent and interpretable, even in spatiotemporal contexts, enhancing the understanding of AI-driven insights and decisions in the realm of urban mobility (Explainable AI).
Mobility AI as a Service
Mobility AI Analytics as a Service (MAIAaaS) is a comprehensive platform designed to facilitate the development, deployment, and management of machine learning (ML) models in the context of urban mobility. It represents a shift from traditional code-centric approaches to data-centric methodologies, emphasizing the integration of data sets, hyperparameter optimization, model testing and validation, precision monitoring, and more. MAIAaaS enables the continuous design, creation, and operation of ML models while accommodating specific needs, such as those found in mobility applications. This service encompasses the entire ML model lifecycle, including continual learning, and offers a catalog of pre-trained models accessible through APIs, tailored to address the unique challenges and use cases within the mobility domain.