Figure 5From: A neural data-driven algorithm for smart sampling in wireless sensor networksApplication of the algorithm to indoor experiments (see Section 2.2.2). (A) Root mean square (RMS) estimation error, and (B) reduction ratio as functions of the uncertainty threshold (assumed proportional to sensor accuracy; the method was run 100 times for each choice of the threshold, mean and standard deviations are shown). The accuracy was assumed to be 0.1°C and 0.3%, for the T and H sensors, respectively. (C) Representative example application for our method: uncertainty of the measurements was assumed to be two times the accuracy.Back to article page