FPGA-Based neural network for accurate distance estimation of elderly falls using WSN in an indoor environment

Abstract

In wireless sensor networks (WSNs), location or distance estimation information for the elderly is a critical issue. This study aims to improve the error in estimated distance between the anchor node and mobile node, which is carried by an elderly person while moving in an indoor environment. The distance between the mobile node and the anchor node was determined based on the measured received signal strength indicator (RSSI) of the anchor nodes and the artificial neural network (ANN). The ANN was implemented into a field programmable gate array (FPGA) so that it could be used in a real-world application. The hardware implementation of the FPGA was executed based on Xilinx (Virtex7). The results revealed that an accurate estimate of the distance error was obtained based on the ANN. The distance error in terms of mean absolute error was 0.019 m for training, 0.07 m for testing, and 0.19 m for validation. In addition, the estimated distances obtained from the FPGA were fully compatible with the actual distances. Moreover, the distance estimation error-based ANN outperformed other existing algorithms and soft computing techniques.