Air pollution is an increasingly concerning issue in urban areas because of its impact on citizens' health. To tackle pollution effectively, accurate monitoring is a must. Precise stations managed by governmental or specialised institutions do exist, but they are both costly and bulky, which limits the potential to deploy them densely. However, recent progress in micro, connected sensors brings new alternative deployment schemes for dense monitoring by low-cost, imprecise sensors. For such a deployment to be relevant relative to urban air quality monitoring aspects, we are concerned with maintaining the system's properties over time. Indeed, one of the major drawbacks of cheap sensors is their drift: chemical properties degrade over time and alter the measurement accuracy. We challenge this issue by designing distributed, online recalibration procedures. We present a simulation framework modelling a mobile wireless sensor network (WSN) and we assess the system's measurement confidence using trust propagation paradigms. As WSN calibrations translate to information exchange between sensors, we also study means of limiting the number of such transmissions by skipping the calibrations deemed least profitable to the system.