J-T. Sung

Abstract: A fault-tolerant classification system in Wireless Sensor Networks (WSNs) has recently been proposed. An adaptive redetection algorithm and an adaptive retransmission scheme were later developed to reduce the misclassification probability of the system when observations of sensors is highly noisy, and the transmission channel between the sensor and the fusion centre of the network is deeply faded, respectively. The observation and the received data are discarded if they are unreliable. However, they still have useful information. This work applies Equal-Gain Combination (EGC) techniques to utilise the unreliable data. Simulation results show that the new adaptive method with EGC outperforms the original one