Authors:
Youmei Hu
1
;
Sijing Duan
2
;
Kuan Li
2
;
Xiaoquan Xu
2
;
Yueqin Wu
2
and
Kun Han
2
Affiliations:
1
College of Advanced Manufacturing Engineering, Chongqing University of Posts and Telecommunications, China
;
2
Chongqing University of Posts and Telecommunications, China
Keyword(s):
Wireless sensor networks, Kalman filter, missing observation, estimation error covariance, stability
Abstract:
Many Wireless sensor networks(WSNs) applications are dependent on clock synchronization technology. The problem of loss of observations for clock synchronization based on Kalman filter estimation is discussed. Firstly, the clock synchronization model of incomplete observation is obtained from the sensor clock reading model. Then, according to the intermittent measurements, Kalman filter formula is deduced and the estimating error covariance recurrence equation is obtained. Considering that the observation loss is random, the statistical convergence of the error covariance is emphatically analyzed. Finally, we show the existence of the critical packet arrival rate, and prove that when the actual packet arrival rate is higher than the critical value, the mean estimation error covariance transitions from unbounded to bounded. Otherwise, we also give the bounds of the covariance of the steady-state error and the boundary of the critical packet arrival rate. Simulation results show the cr
itical packet arrival rate determines the average error covariance transition from unbounded to bounded.
(More)