Towards Optimally Efficient Field Estimation with Threshold-Based Pruning in Real Robotic Sensor Networks

The efficiency of distributed sensor networks depends on an optimal trade-off between the usage of resources and data quality. The work in this paper addresses the problem of optimizing this trade-off in a self-configured distributed robotic sensor network, with respect to a user-defined objective function. We investigate a quadtree network topology and implement a fully distributed threshold-based field estimation algorithm. Simulations with field data as well as real robot experiments are performed, validating our distributed control strategy and evaluating the threshold-based formula for real world scenarios. We propose a theoretical analysis that predicts the system's behavior in real world case studies. The experiments and this prediction show very good correspondence, enabling the accurate employment of the objective function, optimizing the trade-off based on user needs.