This paper addresses the area surveillance problem with a team of multiple aerial robots under communication constraints. In previous work of authors , a decentralized modular architecture was proposed for surveillance missions with a team of homogeneous robots. This paper presents a decentralized decision-making algorithm that solves the problem for heterogeneous aerial robots with different sensing and motion capabilities. The proposed frequency-based approach offers a dynamic and robust solution able to adapt to changes in the area size and robot capabilities, and even to total robot failures. The algorithm can run properly under communication constraints and there is no robot which rules the others. Simulations and experimental results validate the proposed system for heterogeneous robots that can be dynamically added/removed during the execution of the mission.