Consensus-based Kalman filtering algorithm based on compensation strategy for wireless sensor network

In this paper, aiming at the network packet dropout problem, two packet dropout compensation algorithms, the biased compensation method and the balanced compensation method, are put forward to improve the algorithm estimation accuracy based on the consensus-based Kalman filtering algorithm, in which all the node estimations tend to be the same. In the biased compensation method, the undelivered data from neighbours are compensated by the data of its own; and in the balanced compensation method, the weights of all the nodes are updated when packet dropout happens. In this paper, the relationship among the estimation error, algorithm parameters, packet loss rate and the network topology of the two compensation methods is given; and the applicable scopes of the algorithms are determined. Finally, the performances of the consensus-based Kalman filtering, the biased compensation consensus-based Kalman filtering and the balanced compensation consensus-based Kalman filtering algorithms are compared in simulation; and the simulation results show that the two compensation methods proposed in this paper can weaken the negative impact of node failure and packet loss on state estimation, and have better estimation accuracy and robustness.