Distributed Robust Kalman Filtering with Unknown and Noisy Parameters in Sensor Networks

Algorithm 1

1. Initialization: , , , ;

2. while new data is received, do

3. Locally compute the weighted average of estimations and covariances about system parameters:

;

4. Compute the Kalman estimate of the system parameters:

,

,

5. Locally aggregate data and compute the weighted average of estimations and covariances about state estimation:

,

,

,

,

,

where and , is the number of nodes in set ;

if is observable, and for , ,

else

6.Compute the robust Kalman estimate of the target state:

if there exists such that

,

then ;

where,

,

,

,

,

and are updated using the following equation:

,