The use of small sensors that can communicate through wireless links and that are positioned over a wide area can allow to gather precise data about the occurrence of determined phenomena. Depending on the application scenario, the phenomenon to be detected can interest all sensors (global phenomenon (GP)), or just a subset of the deployed sensors (local phenomenon (LP)). One of the most interesting phenomena that can be monitored by wireless sensor networks is fire. In this paper a wireless sensor network configuration for fire detection applications is presented. A chain network is considered where in order to reduce the transmissions and then minimize the power consumptions each node makes a local decision about target absent/present considering its own observation and also the decision made by the previous node. Fusion rules for both the global and local phenomenon scenario are proposed, to minimize the error probability at the last stage of the chain.