This repository contains a small-scale implementation of a Digital Twin (DT) architecture for urban mobility, using FIWARE GEs as core components. The system simulates a real-world bus transportation network in a major Italian city, leveraging Automatic Passenger Counting (APC) systems and GPS data to monitor and predict bus occupancy and movements.
The repository is organized as follows:
- Further Readings contains FurtherReadingList.xlsx with a lists of manuscripts related to the research topic of Digital Twin Architectures, aiming at help readers to understand the context and relevance of our work.
- FIWARE Components contains the docker-compose.yml and .env files to deploy these containers:
- Orion-LD is FIWARE Context Broker. This will receive/send requests in NGSI-LD format.
- IoT Agent for JSON, the bridge that allows devices communication using a simple JSON protocol.
- MongoDB is used by i) the Orion Context Broker to hold context data information, subscriptions and registrations; ii) the IoT Agent to hold device information (device URLs and API keys)
- Timescale timeseries database for persisting historic context.
- Mintaka add-on which services the temporal interface and is responsible for persisting the context.
- Grafana service used for display the persisted time-series data stored in Timescale.
- DataModel Generator contains the Python scripts to interact with FIWARE Components through the ngsildclient library. These scripts aims at showing how the library works to easily interact with FIWARE components deployed in the opportune containers. Due to the fact that data and AI models belong to a private company, no further details can be provided.
- Behavioral Models contains the bpl_predictive_lstm_model.h5 file, representing the predictive model employed for passenger load prediction, and a ReadMe file describing how to create and run the simulation through EclipseSUMO simulator. Further details and files cannot be provided due to data privacy issues.