Achieving availability and reliability in wireless sensor networks applications

Recently, the use of wireless sensor networks has spread to applications areas that are not viable or cost-efficient to be run on other types of networks. The majority of these applications should be run continuously and reliably without interruption. Hence, the two significant quality factors that should be nowadays taken into account in developing wireless sensor networks applications are 'availability' and 'reliability' of application services. The specific characteristics and constraints of wireless sensor networks require a different interpretation of these two factors when developing applications for such networks. In this paper, we propose a middleware layer mechanism for satisfying these two factors. We propose an event-based middleware service that is specifically designed for wireless sensor networks in which a group of sensor nodes forms a cluster and a replicated service is run on each cluster head. The communication model among cluster members and cluster head is based on the publish/subscribe scheme. We show how the replicated services and communication model in cluster nodes increases the availability and reliability of applications running under the proposed middleware.