A lot of technichal and algorithmic problems in the field of sensor networks have been extensively solved. However one essential aspect is still largely unconsidered: Today the implementation of sensor network applications can exclujsively done by sensor network experts. The specific conditions in sensor networks (limitarion of ressources, high failure probability of single devices, etc) demand for special concepts and algorithms. The aim of Aesop's Tale is the  transfer of the concept of Service Oriented Architecture to sensor networks, helping users like biologists or engineers to easily compose their own sensor network applications.

Within the project, IFIS focus on the following topics.

XML: State-of-the-Art tools cannot be used in sensor networks because of their high memory requirements. We want to reduce the memory requirements or even eliminate the XML overhead by embedding XML and XML query languages in widely used programming languages for sensor networks and by a static program analysis.

Continuous queries: Querying sensor networks usually results in long time active in-network queries. These queries can be denoted as continuous queries. Answering continuous sensor network queries requires considering several physical resource constraints, as e.g. limited bandwith, limited supply of energy or limited memory and computational capacity. We want to develop a resource optimized query processing system, which enables XML based declarative querying especially for non experts. By limiting the number of result messages for each continuous query and using distributed index and cache structures, we can further optimize the energy demand of XML in-network sensor query processing systems.

Replication: Providing and executing services in sensor networks reliably means to assure, that a logical existing service survives the failure of a physical node. For this reason it is necessary to  replicate services together with their data continuously. One criterion for the choice of the node, on which replication has to occur, is the remaining battery power and the resulting life expectancy. Another criterion is the spatial distance of nodes, because of the unnecessary increase of communication costs which results in realizing service replication on far away nodes. In this project we design a preferably efficient and resource sparing redundancy infrastructure, that is able to garantee a high degree of reliability.

Transactions: Transactions in sensor networks are a relatively new research area. Former applications in sensor networks needed only possibilities to query and collect measured data while consistent updates of redundantly kept data were not necessary. Recently emerging more complex applications demand a more sophisticated data managment. Considering sensor networks as databases and using replication creates a need for a transaction model analogous to the upcoming database systems in the seventies. In this project we investigate a transaction model specifically designed for the resource constrained sensor networks.

Migratable Web Services: Migratable web services combine approaches of service-oriented architectures, mobile agents and grid-computing. Since they are supporting selforganizing processes and facilitating novel methods for programming sensor network applications, we now plan to transfer this concept on sensor networks. Thus it will be possible for instance to program a new service and to migrate it in an existing network without making modifications on the specific nodes. The service in turn is able to decide on which node it wants to be executed to perform its functionality and to avoid scarcity of resources in the network.

Project Partners

The project partners are the IFIS and the ITM institutes of the University of Lübeck.

Project Leader in IFIS

Prof. Dr. rer. nat. Volker Linnemann

Project Coordinator in IFIS

Dr. rer. nat. Sven Groppe

Researcher

Dipl.-Inf. Jana Neumann

Dipl.-Inf. Nils Höller (Advisor)

Dipl.-Inf. Christoph Reinke (Advisor)

Links 

AESOP'S TALE Homepage