We address the nonuniform coverage of a planar region by a platoon of autonomous mobile agents when communications among them are stochastically intermittent. Given a set of generating points and a suitable metric, the solution of the optimal coverage problem is the well known Voronoi tessellation, with a set of mobile agents converging to the centroids of the corresponding Voronoi cells. In the framework of decentralized motion control, this implementation requires that all agents have knowledge of the state of other agents in the platoon. Here we generalize this scenario by considering the optimal area coverage when a group of agents share information in a time varying, stochastically intermittent fashion. We embed on board of each agent a full state estimator that relies on local estimates and on information received by others, when available. We show that under appropriate conditions on the communication network, all agents' estimates asymptotically converge to true states while maximizing the coverage metric despite intermittent communications. The current work has applications in military and civilian domains including harbor protection, perimeter surveillance, and search and rescue missions. Theoretical results are illustrated through computer simulations.