Forecasting PaaS Design

Description

A platform capable of monitoring and forecasting urban air pollution is designed to tandem with the portable air pollution measurement unit of CAROLINA, as well as other fixed measuring devices. Owing to the lack of available alternatives in the market to enable an accurate estimation based on in-depth analysis and resolution of underlying physics, the strategy adopted has been to develop a new platform. The envisaged platform is expected to involve the following basic common aspects: data acquisition, database, data analysis and post-process, and user interface and experience design. Forecasting platform of CAROLINA will have an additional air quality prediction scheme, which has a potential to fill a gap in the market, achieving an advantageous positioning with respect to the competitors.  For short term predictions functionality, CAROLINA platform, once developed, will be capable of exploring and involving models such as i) microscopic emission model (MOVES) and ii) near real time predictions relying on the comparison and interpolation with a dynamic database with the employment of machine learning capabilities, as well as the possibility of linking with CFD&HPC oriented advanced air quality prediction schemes. Implementation of such sophisticated computational resources to urban air pollution prediction platforms has the potential to be a game-changer in accordance with the anticipated growth of computational power.

Building Blocks and Main Features

Data acquisition	LTE-M modem (GSM) / LoRa-WAN modem (antennas) / MQTT
Database	InfluxDB
Data analysis and post-process	ETL (data keeping) / mathematical proprietary algorithms
User interface and experience	Google Maps API
Air quality prediction scheme	DNS, RANS, LES, Hybrid RANS-LES (TermoFluids, OpenFoam, Ansys Fluent) ; MOVES