(in lingua inglese)
This paper explores an innovative Gaussian mixture state estimation algorithm. By studying the assumptions of prior and posterior pdfs which are based on the quantized innovations, a Gaussian mixture estimator has been derived in the paper. Besides, a recursive posterior CRLB for state estimation using quantized innovations in WSNs is developed. The theoretical lower bound is estimated approximately by adopting a Monte-Carlo method. Performance analysis and simulation experiments show that the Gaussian mixture estimator is better than those quantization KF algorithms.