This paper aims to achieve a balance of power in a group of prosumers, based on a price mechanism, i.e., to steer the difference between the total production and consumption of power to zero. We first set the information network topology such that the prosumers exchange price (power) information with their neighbors according to a chosen information network topology. Based on the exchanged information and the prosumers own measured power demand, each prosumer uses a local control strategy to turn on and off its power generator to cooperatively achieve the global balance. More specifically, the local control strategy results from a distributed model predictive control method based on dual decomposition and sub-gradient iterations. The method achieves a unique dynamic price signal for each prosumer. Simulation results with realistic data validate the method.