Deployment and usage of the portable sensor in GreenIoT project

This thesis planned and executed the deployment of an air quality sensor that was attached on a public transport vehicle.The work was conducted at Uppsala University as a part of the GreenIoT project. The sensor used was a Waspmote Plug & Sense! Smart Environment PRO from Libelium. Vehicle route was chosen among predetermined routes using image analysis on a route map and suggestions from the vehicle providers route coordinators. Local Uppsala city bus route 22 was chosen based on the longest traffic time per day with coverage percentage of 9.91. The collected data was visualized on two different applications based on the Google maps. The sensor measured nitrogen dioxide (NO2), carbon monoxide (CO), temperature, humidity and pressure. The basic sensor code was composed by combining the code snippets provided with the Waspmote IDE. The more advanced functionalities were added to the code after the code worked with the basic functionalities. The sensor communicated with GreenIoT cloud using 4G mobile data. The sensor data was sent to the GreenIoT cloud to be published via the MQTT Broker that would enable the client subscriptions and add the data to the GreenIoT database. The sensor sent successfully 70% of the received air quality measurements and by using the implemented storage functionality managed to send the remaining 30% of the measurements. Roughly 39 000 measurements per sensor probe were recorded. Only 18 records were corrupted to unanalysable form. CO probe measured 38 918 non-zero measurements that equals 99% of the CO data. NO2 probe measured 525 non-zero measurements that equals 1% of the NO2 data. The CO data included high peaks that reached visibly higher than the rest of the measurements. The peaks were analyzed to identify if the measurements were faulty readings or actual pollution levels in Uppsala. Without a controlled environment, it was impossible to say if the peaks were results of a malfunction in the sensor or caused by bad air quality. Data comparison for the CO measurements was made between a second Waspmote sensor and sensor developed by a GreenIoT project partner. Data sets received from the different sensors were restructured for the analysis using Python scripting language. The project sensor measurements were frequently higher compared to the sensor provided by the project partner. The air quality index for CO data was well inside the good air quality scale. A comparison data set for the NO2 data was acquired from the SLB-analys which is a unit within Stockholm City's Environment and Health Administration. The NO2 measurements from the SLB-analys were low and possibly under the accuracy range of the NO2 sensor used. It is hoped that this study could work as a basis for further research on air quality and internet of things in the growing city of uppsala