In this article, we detail and evaluate a distributed algorithm that defines the green lights sequence and duration in a multi-intersection intelligent transportation system (ITS). We expose the architecture of a wireless network of sensors deployed at intersections, which takes local decisions without the help of a central entity. We define an adaptive algorithm, called TAPIOCA (distribuTed and AdaPtive IntersectiOns Control Algorithm), that uses data collected by this sensor network to decide dynamically of the green light sequences, considering three objectives: (i) reducing the users average waiting time while limiting the starvation probability; (ii) selecting in priority the movements that have the best load discharge potential and (iii) synchronizing successive lights, for example to create green waves. Simulation results performed with the SUMO simulator show that TAPIOCA achieves a low average waiting time of vehicles and reacts quickly to traffic load increases, compared to other dynamic strategies and to pre-determined schedules.