In this paper a generalized multiple-model adaptive estimator (GMMAE) is presented that can be used to estimate unknown model and/or filter parameters, such as the noise statistics in designs. The main goal of work provide an increased convergence rate for estimated parameters over traditional (MMAE). Parameter elements generated from quasi-random sequence are drive multiple parallel filters state estimation. current approach focuses on estimating process covariance by sequentially updating...

The recently emerging cubature Kalman filter and the sparse-grid quadrature approximate numerical integrations for mean covariance in Gaussian filters using spherical-radial rule rule, respectively. This technical note reveals that 1) spherical rules can be obtained by projection of rules; 2) third- some fifth-degree directly constructed from rules.

Cubature Kalman filter (CKF) has recently drawn much attention due to its more stable performance than the unscented (UKF). This third-degree cubabure rule based nonlinear may not be accurate enough in many estimation problems. In this paper, a general class of CKFs with arbitrary high-degree accuracy is proposed. It can shown that conventional CKF special case proposed method. A target tracking problem used test filters. will achieve better extended filter, and CKF. addition, it maintain...

A novel measurement update method is presented for quadrature-based Gaussian filters such as the unscented Kalman filters, cubature sparse grid quadrature and Gauss-Hermite filter. Like methods of based on linear minimum mean square error estimation. However, it updates points directly that non-Gaussian structure partially preserved.

This paper presents several extensions of the complex-step approximation to compute numerical derivatives. For first derivatives approach does not suffer substraction cancellation errors as in standard finite-difference approaches. Therefore, since an arbitrarily small step-size can be chosen, method achieve near analytical accuracy. However, for second straight implementation from roundoff errors. cannot chosen. In this we expand upon provide a wider range accuracy both and derivative...

In this paper a generalized multiple-model adaptive estimator is presented that can be used to estimate the unknown noise statistics in filter designs. The assumed unknowns are process covariance elements. Parameter elements generated from quasi-random sequence drive parallel filters for state estimation. current approach focuses on estimating by sequentially updating weights associated with through calculation of likelihood function measurement-minus-estimate residuals, which also...

Amplification distributed denial of service attacks constitute a rapidly evolving threat in the current Internet, which is difficult to be defended for its camouflage and distributability. Inspired by recent advances AI, we consider building an intelligent model that learn defend amplification directly from traffic. Specifically, design novel traffic throttling using reinforcement learning, learning agent makes strategy receiving data. The reward calculated based on proportion legitimate...

In this paper a generalized multiple-model adaptive estimator is presented that can be used to estimate the unknown noise statistics in filter designs. The assumed unknowns are process covariance elements. Parameter elements generated from quasi-random sequence drive parallel filters for state estimation. current approach focuses on estimating by sequentially updating weights associated with through calculation of likelihood function measurement-minus-estimate residuals, which also...

The explosive growth of power data in the process new system business operation has high utilization value, how to maximize value mining at same time, ensure its privacy collection, transmission, aggregation and analysis is a critical issue that needs be addressed. This paper proposes protection scheme (FL-SS) based on federal learning secret sharing. Firstly, framework, information gateway local training model parameters distributed reconfigured security updating while preventing betrayal...

A convex optimization based source estimation method is presented for dynamic models. The effectiveness of the illustrated in context a simple atmospheric puff-based dispersion model. Source process inferring parameters from sensor measurements and physical In dispersion, most important include locations strengths sources as well their number. identification usually involves global search multidimensional parameter space, including large area possible on batch data gathered over reasonably...

In this paper, a new nonlinear filter based on Sparse Gauss-Hermite Quadrature (SGHQ) is proposed for orbit estimation. Although (GHQ) has been widely used in numerical integration, its usage filtering relatively with few successful applications to one-dimensional problems. It difficult use higher dimensional problems because the conventional GHQ that uses product operations implement as number of points increases exponentially dimension. work, we sparse grid method Smolyak’s Product Rule...

Many sensors in chemical, biological, radiological, and nuclear (CBRN) applications only provide very coarse, integer outputs. For example, chemical detectors based on ion mobility sensing typically have a total of eight outputs the form bar readings. Non-Gaussian likelihood functions are involved modeling data fusion those sensors. Under assumption that prior distribution is Gaussian density or can be approximated by density, two methods presented for approximating posterior mean variance....

This article provides a suboptimal approach to the measurement update of state vector and associated error covariance in data assimilation process airborne material dispersion systems, which consists Gaussian puffs sensor measurements local concentrations are bar readings. Based on Bayes rule numerical quadrature techniques, this approximates an interval concentration space with sensor's reading by set discrete points integrals over sums function evaluations at these points. An alternative...

In this paper, the recently developed sparse-grid quadrature filter is compared with cubature Kalman filter. The relation between rule and revealed. It can be shown that arbitrary degree rules obtained by projection of rule. Since both achieve an high accuracy, they are more accurate than conventional third-degree unscented transformation. addition, computationally efficient Gauss-Hermite Monte-Carlo method when used to calculate Gaussian type integrals in nonlinear filtering. comparison...

In this paper, a new nonlinear filter named Salient Point Quadrature Filter (SPQF) using sparse grid method is proposed. The derived the so-called salient points to approximate integrals in Bayesian estimation algorithm. univariate are determined by moment match and then sparse-grid theory used extend point sets multi-dimensional cases. Compared with other point-based methods, accuracy level of can be flexibly controlled algorithm computationally more efficient since number for SPQF...