Distributed data fusion algorithms for inertial network systems

Abstract

New approaches to the development of data fusion algorithms for inertial network systems are described. The aim of this development is to increase the accuracy of estimates of inertial state vectors in all the network nodes, including the navigation states, and also to improve the fault tolerance of inertial network systems. An analysis of distributed inertial sensing models is presented and new distributed data fusion algorithms are developed for inertial network systems. The distributed data fusion algorithm comprises two steps: inertial measurement fusion and state fusion. The inertial measurement fusion allows each node to assimilate all the inertial measurements from an inertial network system, which can improve the performance of inertial sensor failure detection and isolation algorithms by providing more information. The state fusion further increases the accuracy and enhances the integrity of the local inertial states and navigation state estimates. The simulation results show that the two-step fusion procedure overcomes the disadvantages of traditional inertial sensor alignment procedures. The slave inertial nodes can be accurately aligned to the master node.