Transfering the paradigm of service-oriented architecture (SOA) to sensor networks became an important research area in the last years. Amongst others, a couple of approaches deal with the flexible and robust service discovery and usage of services which takes the unsteady nature of WSNs into account. However, the approaches do not consider the case that services may get unavailable during their activity phase because of node failures. In this case, already collected or computed data will get lost and the service request of the service consumer remains unanswered. In this paper, we propose a redundancy infrastructure for service-oriented WSNs which deals with this special problem. The framework consists of an adaptive data replication technique and a service recovery solution. Whereas the replication technique ensures the survival beyond node failures, the recovery protocol restarts a failed service using the replicated data and enables the service to continue its functionality.

XML is the defacto standard for data exchange applications like those in the WWW. However, due to the limited hardware resources, wireless sensor networks abstain from using verbose data formats like XML. Nevertheless, XML as a standardized data exchange format in wireless sensor networks is a means to support more complex data management and heterogeneous networks. Moreover, XML is a key feature towards service-oriented sensor networks that exchange structured information by using SOAP. Recent work has shown that XML can be compressed to meet the general hardware restrictions of sensor nodes while still supporting updates. In this work we outline the vision and benefits of XML usage in wireless sensor networks, show how to evaluate XML queries in wireless sensor networks and how query results can be compressed to lower the comunication overhead. We therefore present an XPath engine on updateable compressed XML data for sensor nodes and an experimental evaluation showing that the performance of our XPath engine fulfills the requirements of today's applications even on sensor nodes.

Due to the highly distributed nature and special basic conditions such as limited resources, implementing and maintaining a sensor network application is a tedious task. In this demonstration we present a service-oriented operating system for sensor nodes and a framework that allows a composition of services on the nodes. We show how sensor network applications can be composed and modified by using different services.

In this demonstration we show how a sensor network application can be composed by different services. We demonstrate how the services dynamically distribute within the network and how this distribution automatically adapts to changes in topology or user demands.