Wireless sensor networks constitute a novel measuring technology which enables the monitoring of large environments at a high resolution. Current application areas range from environmental monitoring to industrial maintenance. Such networks consist of small, battery-powered nodes equipped with sensors and constrained computation and memory resources.

Basically, sensor networks are deployed to gain information concerning a particular environment/ object based on measurement data (e.g. humidity, temperature). This requires the acquisition of the relevant sensor readings from the network, i.e. acquiring current measurements focussed on those aspects that are of interest to the user. From a data processing perspective this corresponds to filtering and combining the data.

However, acquiring the data and focussing the data to relevant aspects turns out to be a severe hurdle for users. Declarative queries are a well-established technology capable of overcoming this problem. It simplifies the interaction with sensor networks since it enables a user to specify what information he is interested in without having to state how the data is acquired. This latter task is part of the query processing.

The underlying idea which enables declarative queries against sensor networks is to regard the network as a database-like relation, the attributes of which correspond to the sensors (e.g. humidity, temperature, etc.). In this setting, the goal of the query processing is to answer queries “in a good manner”. Since the nodes are typically battery-powered, minimizing the energy consumption is the most important optimization goal. Most notably, since communication is the most expensive operation of a node by orders of magnitude, minimizing the communication costs is of utmost importance.

With respect to query processing, efficiently executing simple operators such as selection and projection as well as aggregation has been studied extensively. In contrast, efficiently processing complex queries like join queries or queries that require combining the data of major parts of the network is an open problem. For instance, the join enables to relate data from different nodes to each other and thus is often part of a filtering/ data processing.

For such complex queries, currently the only way of processing them is to consolidate the data from the entire network at the base station and to perform a centralized processing. However, this is the worst case solution with respect to the communication costs. The goal of this project is to efficiently process such complex queries.