| In order to adapt to the informatization of war, complex environmentand gathering more information of the target, more and more sensors of manytypes are under using. Limited by the complexity of the computation, re-al-time requirement, and the use of places, it is necessary to optimize thestructures and algorithms of the multi-sensor system. When the sensor cancommunicate with its neighbors, the traditional distributed filter with con-sensus algorithm can be used to achieve the target tracking and reach a con-sistent state of the entire sensors. The research work in this paper is concen-trated on three aspects as follows:First, the basics of consensus algorithm is studied, including graph the-ory, algorithm theory analysis, the convergence conditions, and the consensuspreformation. The consensus algorithm is introduced into the traditionalKalman filter, involving design of the information interaction processing al-gorithm with a two-fold objective:1) estimate the state of the target of inter-est and2) reach a consensus on the state estimate.Secondly, the target tracking algorithm based on consensus is studied. Inorder to improve the target estimation accuracy, enhance the real-time pre-formation, and reduce the impact on the overall of the estimation error, thispaper presents optimal distributed information fusion criterion to optimize the Kalman filter based on consensus. At last, an improved weighted Kalmanfilter based on consensus is proposed.Finally, study on tracking algorithm of target in clutter based on con-sensus. In complex environment the sensors scan more effective measure-ments, leading to pay greater communication cost when a measurement in-teracting between neighbor sensors, so this is not conducive to real-time tar-get tracking. In order to reduce communication cost, simplify algorithm, wecan combine the data association technology and Kalman filtering techniquebased on consensus and the new algorithm only requires the interaction of thepredicted state estimation. |
PDF Full Text Request