Project title: Software Sensor Augmentation at Environmental Data Analysis Laboratory
Acronym: SSA@EDAL
Project identifier: UIP-2019-04-1737
Type of project: Installation Research Project
Funding body: Croatian science foundation (HRZZ)
Project duration: February 2020 – February 2025
Summary
Sensors which provide information about environment do not necessary have to provide measurement of natural phenomenon. Many data comes from (software or hardware) sensors that measure social component in a social environment – from widely used mobile devices to special purpose devices. With the rise of Internet of things (IoT), data reconstruction is even more in focus. A similar trend can be noticed in the development of Intelligent Transport Systems (ITS), i.e. with the growth of the number but also the types of traffic data sources that can now be: inductive loops, mobile sensors, GPS, video cameras, Bluetooth vehicle tracking. Often it is difficult to cover the large area with sensors, and sometimes if the sensor has large coverage the time between acquisition of two samples can be undesirable long, thus a certain trade-off between spatial and temporal resolution has to be done. Two extreme cases can be observed in oceanography where e.g. the local measurements have poor spatial resolution, but good temporal resolution, and the satellite measurements have good spatial but poor temporal resolution.
In such scenarios a software sensor can be utilized to mimic the measurement of the hardware sensor at that specific time and location. In order to accomplish this we aim to explore the regularly occurring patterns in the observed systems, and – with the use of machine learning – model the system in order to understand the main concepts that drive the characteristic behavior. The knowledge represented in the model may then be used to reconstruct data at times or locations for which the data are missing. The aim is to utilize the methodology in at least two different scenarios – in oceanography and ITS – in order to explore the possibility of transferring the knowledge between different problem domains.