Research On Collaborative Location Algorithm Based On Fisher Information

Location information is critical for many future and current wireless applications.As a novel positioning method,cooperative positioning algorithm provides a broad prospect for wireless positioning.Different from traditional positioning,the cooperative positioning method uses not only the measurement from the anchor and the relative measurement between the unknown nodes.Cooperative positioning can not only improve the positioning accuracy but also complete the positioning in the situation where the traditional method cannot achieve the positioning.So,this paper studied on the implementation of cooperative positioning algorithm and the problem of cooperative node selection.The main contents of this paper are as follows:Because of the complexity,the joint posterior probability density function of the network nodes is not conducive to the calculation.Factor graph combined with the sumproduct algorithm can realize the marginalization of the joint distribution of the network nodes.However,the positioning error of message passing algorithm based on the parameter is large in the non-Gaussian,non-linear system.In order to realize the message passing of the sum-product algorithm,this paper analyzed the drawbacks of the message passing algorithm based on the parameter and the characteristics of the cooperative network factor graph.In this paper,an algorithm is proposed to implement the message passing through particles.The proposed message passing algorithm use importance sampling and Gibbs sampling to obtain the sum and product of message respectively.The simulation results show that the particle method improves the positioning accuracy in the nonlinear and non-Gaussian environment.The key of cooperative positioning algorithm is quantifying the uncertainty of cooperative information.According to the theory of network navigation,cooperative information from the neighboring unknown nodes and neighboring anchors all can be quantified by the fisher information matrix.Cooperative information quantified by fisher information matrix not only can be used to analyze the factors that affect the positioning accuracy easily and to calculate the positioning error directly.The positioning node need receive the neighbor information during the cooperation.In order to minimize the information redundancy between nodes,the node selection is necessary.Cooperative information quantified using Fisher information can represent the relationship between positioning errors and cooperative information.Therefore,this paper uses fisher information for neighbor selection,so that the positioning error is the smallest after the user node cooperate with the selected neighbors.The simulation results show that the node selection algorithm can select the neighboring nodes which are most beneficial to the user node.