RSSI-based positioning for health care service using artificial neural network approach

The fluctuation of received signal strength indicator emerges poor accuracy in healthcare location monitoring service. To enhance the capability of position classification, this paper represents positioning system based on ZigBee standard using artificial neural networks algorithm. Time-delay Multi-Layer Perceptron is proposed by using Levenberg-Marquardt optimization. For the result, the average error of four empirical experiments reaches to 7 centimeters with 1.5 square meters grid resolution. The reduction of grid scale in order to extend output resolution is also a limitation for RSSI-based positioning due to an uncertainly and ambiguity of RSSI vector.