Air Quality pilot with Hopu

Pilots in Mechelen & Bruges


How can we monitor and improve air quality in our cities using open data?


  • Project: Air quality
  • Location: Mechelen (Belgium, 86.600 inhabitants)  and Bruges (Belgium, 118.600 inhabitants)
  • Challenge: How can we monitor and improve air quality in our city using open data?
  • Solution: Sensors deployed on strategic points around the city



Good air quality in the city is important for a number of reasons. First of all for the health of our citizens, who may suffer from lung, heart and other related problems due to poor air quality. Furthermore smog and acid rain can have an effect on the environment and infrastructure. A number of circumstances such as time, weather, traffic, events or street layout can have an effect on the air quality. It is therefore imperative to measure the air quality accurately throughout the city to be able to locate problem areas and to make the right decisions to improve it.



Air quality sensors were placed in seven locations in different circumstances to find differences in concentrations with a lot of traffic, green spaces in the city and rural area next to the city. These sensors tracked multiple gases (NO2, O3, CO en SO2), particulate matter, temperature, humidity and noise levels. Also a weather station was placed to find correlations with (high) temperature, rainfall or wind. Those sensors continuously sent data to a platform every minute. The data was published as real time open data following the Fiware data models. It was presented in dashboards for consultation by interested city departments and citizens, and also machine-readable via Orion Context broker.



  • There was a big delay in the delivery of the sensors due to unforeseen production issues and a limited production and testing capacity.
  • Transport of the sensors caused broken parts in the sensors. Either the components should be made robust enough to survive international travel, or special care transport needs to be done.
  • Installing devices in the public domain, requiring power is not straightforward. Certainly for a PoC, locations where these devices could be installed without a lot of extra agreements for instance were not always the best locations to get the most important data regarding air quality. The PoC is thus more a test of device operation and less a small air quality assessment.
  • After the installation the sensors failed often due to power supply issues, right now only one of them is still working properly. It is therefore important not to underestimate installment issues (power, location, reachability). Half of the sensors were installed on lamp posts. Firstly this implies the necessity of batteries in the devices, and secondly the electricity profile caused distortions of the MCU resulting in malfunctioning and even nonfunctioning.
  • Mobile connectivity is a critical component. It is therefore important to know/decide beforehand what network you will use to send the data of the sensors. 2G/4G didn’t always work as expected and Wifi or LAN is not available everywhere. LoRa is a better alternative, but is expensive if you send a lot of data and often.
  • The Spanish manufacturer of the sensors had no experience regarding mobile connectivity in our region, which lead to problems. You should therefore aim for applicants that do have that kind of experience.
  • There were also problems with the AI for the processing of the gas measurements.
  • Data was sent every minute, but this is not a frequency set in stone. Consider the use you will make of this data to set the frequency.


Doing this pilot revealed a number of not-so-obvious factors to take into account when considering implementing a sensor network:

  • Fixing the devices in the public domain might mean:

Fixing them on private property

Using power of a private owner

Providing extra power connection

  • Connectivity is a crucial element. So either make sure you are responsible for the network quality, or that the provider has enough experience with the connectivity platform in the country where the devices are installed.
  • Start with renting the devices, certainly when they are not fully market-ready. This will avoid a large cost that might be obsolete if the PoC proves unsuccessful.
  • It’s a difficult equilibrium between testing innovative technology (experimentation) and implementing a proven technology.


We really wanted to gather data at least six months to have a big enough and relevant dataset. Due to the delays and issues, the pilot is still ongoing. It remains to be seen whether the pilot will continue or will be stopped.