Radar selection for single-target tracking in radar networks

A sensor networks comprise a large number of sensor nodes collaboratively collecting information, carrying out simple computation with their onboard processing capabilities. In recent years, sensor selection has become an important problem in managing sensor networks. Facts have shown that effective sensor selection benefits much to the performance of the sensor networks, such as target tracking. Among many works about sensor selection problems, a criterion to select sensors based on information theory is proposed in a recent paper. However the criterion in their work is suitable for only some special cases because of the stringent constraints. In this paper, we propose a novel general criterion with the consideration of correlation in the measurement errors from different radars. Based on theoretical analysis we define a matrix norm (∥R₁ + R₂∥₂) comprised of different measurement error covariance matrices as the measure to select radars. Meanwhile, Kalman filter combined with Covariance-Intersection (CI) information fusion is adopted to deal with the tracking process considering the measurement errors are correlated or not. Simulations have shown that the criterion proposed in this paper is applicable for a more general sensor networks and the tracking performance is also improved with the consideration of the correlated observations from different resources.