Distributed decision-making for sensor networks

The development of reliable autonomous decision-making in sensor networks requires robust methods

of assessing the output of surveillance systems and determining actions without human intervention. This is particularly crucial for large-scale multi-sensor systems where the involvement of an operator to make the decision on how to allocate resources is infeasible. The objective of this proposal is to develop key performance measures for assessment and analysis of multi-sensor multi-target tracking systems.

The result of these developments will inform the development of a new robust class of information-based sensor-scheduling algorithms to enable intelligent autonomous decision-making in large-scale sensor networks. The focus is on the development of basic tools in estimation theory that are fundamental for

statistical analysis applied to the specific challenges of multi-target surveillance where there is uncertainty in the number of object and the object states. These developments will inform the assessment of scalable large-scale tracking systems from many distributed sensors.