Estimation of Observation Error Probability in Wireless Sensor Networks

In this letter, we first of all propose for a parallel wireless sensor network (WSN) a decoding technique that well exploits the correlation knowledge of the sensing data to be transmitted from each sensor to the fusion center (FC). This letter then derives an algorithm to estimate the observation error probabilities, representing the correlation, of the links between the sensing object and sensors. The convergence of the algorithm is also evaluated. Furthermore, the comparison of bit-error-rate (BER) performance between two cases, one uses estimated observation error probabilities, the other assumes the full knowledge of the observation error probabilities, is made. The simulation results show that the difference is only around 0.3 - 0.5 dB in per-link signal-to-noise power ratio (SNR), depending on the number of sensors