Recently, air pollution monitoring emerges as a major issue of the development of smart cities and the well-being of citizens. Air pollution is traditionally monitored using some measuring stations that are accurate but expensive, big and inflexible. This leads to bad global estimations of pollution concentrations. The emergence of air quality sensors, which are less expensive, allows to consider a new pollution monitoring paradigm based on the wireless interconnection between these sensors. This allows to ensure a district-wide air pollution monitoring. In this paper, we tackle the minimum-cost node positioning issue for the detection of air pollution thresholds. We propose two models for wireless sensors deployment while taking into account the air pollution modelling and the probabilistic sensing of nodes. We evaluate our deployment models on a real data set of Greater London and conduct extensive simulations to study the impact of some parameters, among which sensors' height. Results show that the deployment cost depends on the dispersion of pollutants in the area of interest and can be minimized by placing sensors at a height close to the one of pollution sources.