Research On Sensor Management Of Target Tracking Based On Bearing Measurement Model

With the rapid development of microelectronic mechanical system technology,the application of wireless sensor networks is becoming more and more extensive.Target tracking is a typical application scenario of wireless sensor networks.However,continuous activation of a large number of sensor nodes to participate in the target tracking process will inevitably cause excessive energy consumption of the sensor itself,which will affect the life of the entire network,especially the energy and bandwidth of the wireless sensor network.It is usually limited,so how to allocate sensor resources scientifically and reasonably while ensuring the accuracy of target tracking is very important.At present,target tracking sensor management has become a hot and frontier research topic at home and abroad.It is of very important scientific significance to study the corresponding sensor management algorithms for different target tracking scenarios.In this paper,the sensor management research is carried out on the target tracking application scenario under the classic passive detection model-the azimuth measurement model.The main research work is as follows:(1)Aiming at the problem of how to select a better subset of sensor nodes as task nodes active to participate in target tracking in a single target scenario,the performance indicators in the target tracking sensor management method are studied,and conditional posterior Cramer-Rao lower bound is selected,The judgment criterion of sensor management problem is based on the study of its basic theory and mathematical modeling process,deduces the analytical expression form of the condition posterior Cramer-Rao lower bound under the bearing measurement model,and constructs a kind sensor management objective function based on this conditional posterior Cramer-Rao lower bound,and finally compared with the mutual information method and the nearest neighbor method respectively.The experimental results show that the method in this paper can select a better subset of sensor nodes and obtain more accurate target state estimation results.,And the calculation efficiency is better.(2)Solving the objective function of sensor management is an NP-hard problem.The amount of calculation using the exhaustive method will increase exponentially with the increase of the number of sensor nodes.Therefore,this paper proposes an improved hybrid binary whale optimization algorithm to solve sensor management problem.The algorithm uses a nonlinear adaptive convergence factor and an exponential function-based dynamic perturbation weight to effectively balance the search and development process of the algorithm with each other.It has strong global optimization capabilities and rapid convergence capabilities,and can effectively avoid falling into local optima.In addition,through the performance analysis of different conversion functions when dealing with different optimization problems,the algorithm adopts a new V-shaped conversion function to realize the position mapping in the binary space,so as to adapt to the sensor management problem in the tracking scenario of this article.Finally,the performance of the algorithm proposed in this paper is compared with the exhaustive method and other swarm intelligence optimization algorithms when solving the above objective function through experiments.The results show that the algorithm proposed in this paper performs better in the optimization and convergence capabilities,and can quickly find a better subset of sensors to obtain better tracking accuracy.(3)Aiming at the problem of competition between targets with different threat levels and sensor resources in the multi-sensor multi-target tracking scenario,based on the sequential Monte Carlo implementation of the cardinality balance multi-target multi-Bernoulli filter,this paper proposes a new method based on Threat degree combination weighting sensor management method,this method fully considers the factors that affect the target threat degree level,comprehensively evaluates the threat degree from three aspects: target movement situation,target detection probability and target existence probability,and first obtains through the Analytic Hierarchy Process subjective weight,while obtaining objective weight through the Entropy Weight Method,and then use the Kullback-Leibler divergence to evaluate the degree of dispersion between subjective weight and objective weight,and calculate the optimal weight corresponding to different factors.Determine the maximum threat target and its predicted probability density according to the results of the threat evaluation calculation,and use the information gain of the maximum threat target as the sensor management objective function for calculation,and select the optimal subset of sensor nodes for sequential fusion of measurement data to realize the sensor management of multi-sensor multi-target tracking.The experimental results show that the method proposed in this paper can effectively select the subset of sensor nodes that optimize the tracking performance of the most threatening target,and is more suitable for multi-target tracking scenarios that require prioritization.