A MLE Based Algorithm for Registration in Sensor Networks

Sensor bias may severely affect the use of measurements acquired in sensor networks. A registration algorithm based on maximum likelihood estimate (MLE) for estimating the bias is presented in this paper. Firstly, the impact of residual bias on target state estimate was analyzed. Then the likelihood function with respect to residual sensor bias was derived, and an iterated algorithm was presented to obtain MLEs of bias. The presented algorithm is a batch algorithm. Simulations are performed to demonstrate the effectiveness of the presented algorithms.