Most of the current in-network data processing algorithms are modified regression techniques like multidimensional data series analysis. In our opinion, several algorithms developed within the artificial neural-networks tradition can be easily adopted to wireless sensor network platforms and will meet the requirements for sensor networks like: simple parallel-distributed computation, distributed storage, data robustness and auto-classification of sensor readings. Lower communication costs and energy savings can be obtained as a consequence of the dimensionality reduction achieved by the neural-networks clustering algorithms.

In this paper we will present three possible implementations of the ART and FuzzyART neural-networks algorithms, which are unsupervised learning methods for categorization of the sensory inputs. They are tested on a data obtained from a set of several motes, equipped with several sensors each. Results from simulations of deliberately made faulty sensors show the data robustness of these architectures.