Poor air quality in urban environments is one of the major problems of modern cities. This paper presents a crowd-sensing approach that leverages the mobility of public transportation to monitor air quality in different inner-city areas over the course of a day. To meet this objective, we present a system architecture for the compilation of environmental data using low-cost mobile sensor devices equipped in public transport vehicles. Thus, the network topology has been designed to be robust and reliable, tackling connectivity problems through the use of delay-tolerant networks and Bluetooth ad-hoc gateways, and to be flexible, dealing with changing topologies through the use of mobile ad-hoc networks. The tests carried out in the public transportation of Cartagena (Spain) show the feasibility of the proposed architecture for this vehicle to infrastructure service and the effectiveness of deploying a mobile measurement system instead of fixed stations, which is the usual solution for many smart cities.