by HOP Ubiquitous S.L.
Summary
Lately, cities activities are increasing, the more people are, the more are pollution we can find in our streets and roads due to, for example, people usage of different kinds of motor transport. This have increases the need of study this phenomena, in order to find a solution suitable from everyone, a solution which is easy and scalable.
Environmental monitoring is the basis in a Smart City infrastructure to obtain a high level of awareness about the impact of urbanization, mobility and industrialization. For that purpose, smart cities are deploying gases sensors (NO2, SO2, O3, CO) as recommended by the World Health Organization (WHO) to determinate the air quality index.
Unfortunately, air quality and gases sensors have a complex behaviour based on electrochemical reactions, which require a calibration and tuning process to provide an accurate value, at the same time, even when they are calibrated in laboratory, chemical material is sensitivity to multiple gases (cross-sensitivity) making it not very selective. In addition, sensors lose their sensitivity and accuracy after six months and they are totally considered useless for monitoring after 2 years (maximum lifetime). Therefore, it presents a high maintenance cost and also a big challenge to guarantee its sustainability in long term.
Recent studies have demonstrated a correlation among the different gases concentration for every city; these values can be calculated in order to compensate cross-sensitivity. These algorithms and relationships among gases will enable maintenance/tuning of sensors during time, taking into account correlations, cross-validation and region characterization. Additionally, it is required to obtain some parameters (meta-data) about quality of data, in order to avoid wrong decisions or misunderstandings based on inaccuracies coming from sensors, therefor it is necessary the identification of sensors misbehaving in order to discard the data to mitigate errors, and in the best cases to recalibrate them (self-healing) in order to recover the system to obtain good values for the future.
FineTune aims to establish these algorithms/correlations among gases in Crete and Santander based on the historical data, in order to use them in new deployments. For this purpose, it has been used the real data from existing deployments from SmartSantander and FINE Testbeds.
FineTune has developed algorithms/correlations to validate air quality sensors, and it has also defined a holistic approach to the usual calibration approach based on the evolution of the sensors and cross-validation among the different sensors, gases correlation and cross-sensitivity.
FineTune has been able to identify all the performances, metrics, data quality from all the sensors in Crete and Santander, at the same time that it has been able to correlate and obtain relevant insights about gases sensors behaviours and evolution.
FIESTA-IoT has been crucial for this experiment, since it provides the homogenization of the data coming from the different testbeds from FIESTA-IoT in the vertical of Smart Cities, and it has enabled us the opportunity to evaluate several conclusions and contrast results from a location in other location.
After FIESTA experimentation, now we have the baseline of knowledge for air quality sensors calibration which is being continued through a deployment of a laboratory for air quality sensors calibration including Mass Flow Controller, incubators and different air concentrations generation; we have found that a proper calibration requires of a reference system with a high accuracy in order to be able to build the models that define each sensor; since they are electrochemical sensors every system is different and every system requires a full modelling and calibration process. For that reason, it is required to have a stable environment where understand the differences and offsets among the different sensors.
Finally, we will continue this research line as a key part for HOP Ubiquitous portfolio (in particular for Smart spot product extensions for air quality sensors), in order to define new techniques and methods to calibrate sensors and evaluate their accuracy. In particular, we are defining a new calibration methodology which is being patented.