Global goal of this research group is to gain fundamental insights in the design, construction and the adaptive operation of resource-scarce, heterogeneous, intelligent networks consisting of static and mobile sensors. The interdisciplinary group of researchers set of biologists, computer scientists and electrical engineers aims to develop and demonstrate cutting-edge methods and techniques for gathering data, data management and data analysis. The newly developed methods and techniques should exemplary be deployed for behavior research of bats, a species that is classified as deserving special protection by the European Union. Concretely, bats should be equipped with miniaturized, ultra-light weight radio transmitters, which also have limited memory and computing capacity for data processing.
A network of ground-based, static sensors is used to monitor a variety of bats equipped in such a manner. Figure 1 illustrates the scenario from the perspective of biological application. In behavioral studies on bats usually relatively coarse, i.e. a few meters accurate estimates of the flown trajectories without real-time requirements are sufficient. As in many applications, therefore, the energy and cost input that comes along with positioning with sub-meter precision and real-time capability does not pay off. Only occasionally stationary or time-dependent precise location is of interest to study the social or the hunting behavior more accurately. Primary goals in the design of the sensor network are therefore spatial scalability of network size and minimal weight, minimal energy consumption, minimum volume and bat-friendly form factor of the mobile sensor nodes. Scalability in location accuracy in the interplay with the life-time or duration of the recording of the mobile sensor nodes is a secondary optimization objective, which is to be prepared in the first phase of the project, but investigated closely only in a second phase of the project.
The outlined approach implies that massively distributed storage and computing resources must be identified and suitably used. The network of ground-based computing nodes with varying characteristics, which interconnect in a self-organizing manner, is to allow capturing the movements of the bat-sensor nodes temporally referenced and evaluate them tailored to biological questions as automatically as possible. Due to the scarcity of resources in terms of energy, storage capacity and computing power, and to support variable, requirement adapted data stream requests, hardware and software as well as the applied protocols are designed as energy-aware and adaptive. Intended to optimize the energy management a data pre-processing on mobile sensor nodes and between bats and to the ground-based sensor network a, albeit highly restricted data transfer should be possible.
These objectives throw a bunch of complex scientific issues in terms of the topology of the network, including room layout with directional antennas, in terms of overall system architecture, in terms of programming models to support the user in terms of an energy-aware, resource-saving operating system on the mobile node, in terms of handling of data stream queries to the network, in terms of data management and communication in the network, in terms of the used measuring methods of obtaining location information, in terms of effective data fusion methods for trajectory estimation and not least in terms of miniaturization and design of mobile wireless nodes. These modeling and design tasks are derived from the requirements arising from the scientific questions of biologists. The individual questions are discussed in detail in the context of the partial project descriptions.
In eight coordinated subprojects energy-efficient chips are developed, energy-efficient mobile sensor nodes are built, the ground-based sensor network for tracking is designed, phase-coherent measurement methods for accuracy enhancing localization are used, communication, data and network management for "spontaneous" reprogrammable sensor network nodes is provided, data stream requests across systems are optimized, software infrastructures for sensor networks and limited resources are explored and not least the habitat use of bat communities is analyzed. The ninth sub-project, the aspect-oriented simulation of the overall system, plays a key role, since all processing levels should be system-wide integrated and optimized. In this way, a powerful tool for behavioral research will emerge, which is composed of technology building blocks that can be easily modified with a view to other applications and integrated to a whole new system.