Fault recovery in single-hop sensor networks

Wireless ad hoc Sensor networks are gaining increasing attention as they are suitable to monitor a wide variety of environments, where the deployment of classical wired communication infrastructure is neither economical, nor possible. The network collects environmental information which, in turn, is sent to a sink node (accessible to external users) when the connection with the sink node is available. If a sensor fails when the connection with the sink node is unavailable, all data sensed by this sensor are definitively lost. Neither the classical approaches to fault recovery nor the recently proposed approaches for the recovery in mobile computing are appropriate under this scenario, because they would require that the sensors be equipped with stable storage and/or a permanent connection with the sink node. Such requirement can hardly be met in sensor networks. This paper introduces a fault recovery scheme for single hop wireless sensor networks, where the network is able to operate independently of the sink node for long periods of time. The proposed scheme does neither require that the sensors be equipped with stable storage nor that they are permanently connected to the sink node. Rather, the sensors cooperate to maintain redundant information in their memories in order to be able to recover the lost information after a failure, and to distribute the recovered information among the memories of the surviving sensors. The proposed scheme minimizes the memory overhead.