Adaptive Neural Network Controller for Modeling Link Quality in WSANs

Abstract

paper develops a new approach based on off-line trained artificial neural networks (ANNs) for an adaptive controller to compensate the packet loss and delay which almost occur in the channels of wireless sensor/actuator networks (WSANs). A model reference adaptive controller scheme is used to enhance the performance of the control system. The capability of training of ANNs is proposed to create a new controller which can be called model reference adaptive neural network controller (MRANNC) in WSANs to improve the link reliability and make the overall WSAN performance better. The dynamics of the plant are stored in the ANN. The output of the plant is regulated according to a prespecified reference signal. The simulation results show a promising approach that can be used to enhance the link quality in wireless control system. The packet loss of channel can be considerably adapted at different sampling times with and without time delays.