Distributed Lattice Kalman Filtering

The traditional single target real-time positioning accuracy is low and the multi-target centralized positioning requires information fusion center and large communication volume. In the nonlinear Kalman filter frame-work, the target groups are divided into clusters according to the link topology. The distributed lattice Kalman filter (DLKF), which is a quasi-Monte Carlo method (QMC) based on lattice rules, was derived by splitting and analyzing the state information obtained by sensors configured on cluster targets. The proposed lattice Kalman filter (LKF) method is based on Cranley-Patterson shift method and uses Korobov lattice rules to determine the low-difference sample points that generate random shifts. Then, in the distributed network system, the weighted rules are updated based on the weighted average consistency (WAC) algorithm to improve the precision of the fusion results within a limited consensus step. The test inferences evidence that the explored approach requires fewer sampling points and reduces the difficulty of integration, which is the most important core factor that DLKF is superior to other filters.