Hongbing Ji

Extensions of the SMC-PHD filters for jump Markov systems

The probability hypothesis density (PHD) filter is a promising algorithm for multitarget tracking, which can be extended for jump Markov systems (JMS). Since the existing multiple model sequential Monte Carlo PHD (MM SMC-PHD) filter is not interacting, two extensions of the SMC-PHD filters are developed in this paper. The interacting multiple-model (IMM) SMC-PHD filter approximates the model conditional PHD of target states by particles, and performs the interaction by resampling without any a priori assumption of the noise. The IMM Rao-Blackwellized particle (RBP) PHD filter uses the idea of Rao-Blackwellized to further enhance the performance of target state estimation for JMS with mixed linear/nonlinear state space models. The simulation results show that the proposed algorithms have better performances than the existing MM SMC-PHD filter in terms of state filtering and target number estimation.

A novel track maintenance algorithm for PHD/CPHD filter

Probability hypothesis density (PHD) filter and cardinalized PHD (CPHD) filter have proved to be promising algorithms for tracking an unknown number of targets in real time. However, they do not provide the identities of the individual estimated targets, so the target tracks cannot be obtained. To solve this problem, we propose a new track maintenance algorithm based on the cross entropy (CE) technique. Firstly, the particle filter PHD (PF-PHD) algorithm is used to estimate the target states and the target number. Then, the results of the estimation are used as vertexes to construct a connectivity graph with associated weights, and the CE technique is employed as a global optimization scheme to calculate the optimal feasible associated events. Furthermore, due to the advantages of the CPHD filter and the Rao-Blackwellized particle filter (RBPF), we propose another track maintenance algorithm based on the CE technique, named the RBPF-CPHD tracker, which can further improve the track maintenance performance due to the more accurate state estimates and their number estimates. Simulation results show that the proposed algorithms can effectively achieve the track continuity, with stronger robustness and greater anti-jamming capability.

A novel fast partitioning algorithm for extended target tracking using a Gaussian mixture PHD filter

In an extended target PHD filter, the exact filter requires all possible partitions of the current measurement set for updating, which is computationally intractable. In order to limit the number of partitions, a fast partitioning algorithm for extended target Gaussian mixture PHD (ET-GM-PHD) filter is proposed, which substitutes Distance Partitioning with a fuzzy ART model. Alternative partitions of the measurement set are generated by the different vigilance values in ART. Suitable measures and remedies are given to handle the problems arisen by overestimation of target number and spatially close targets. The simulation results show that the proposed algorithm can well handle the close-spaced targets and obviously reduce computational burden without losing tracking performance, which implies good application prospects for the real-time extended target tracking system.

Radar emitter recognition based on cyclostationary signatures and sequential iterative least-square estimation

Radar emitter recognition plays an important role in electronic warfare (EW). Specific radar emitter recognition is the state-of-art technology of emitter recognition, which can recognize the different radar devices of the same type. It is a composite task that involves radar signal interception, modulation recognition, features extraction and classification. In this paper, first we study the unintentional modulation on pulse (UMOP) features of radar emitter. Then the iterative least-square method is introduced for estimation of the UMOP features. Because of the discriminatory capability and abundant information of cyclostationary signatures, the zero frequency slice of cyclic spectrum is used for specific radar emitter recognition. Based on these, the sequential iterative least-square (SILS) algorithm is proposed for the online recognition of radar emitters. Finally experiments on three simulation radars and eight actual intercepted radars with the same type verify the correctness and validity of the proposed method.

Gaussian mixture reduction based on fuzzy ART for extended target tracking

This paper presents a global Gaussian mixture (GM) reduction algorithm via clustering for extended target tracking in clutter. The proposed global clustering algorithm is obtained by combining a fuzzy Adaptive Resonance Theory (ART) neural network architecture with the weighted Kullback-Leibler (KL) difference which describes discrimination of one component from another. Therefore, we call the proposed algorithm as ART-KL clustering (ART-KL-C) in the paper. The weighted KL difference is used as a category choice function of ART-KL-C, derived by considering both the KL divergence between two components and their weights. The performance of ART-KL-C is evaluated by the normalized integrated squared distance (NISD) measure, which describes the deviation between the original and reduced GM. The proposed algorithm is tested on both one-dimensional and four-dimensional simulation examples, and the results show that the proposed algorithm can more accurately approximate the original mixture and is useful in extended target tracking. The paper presents a global GM reduction algorithm, ART-KL-C, via clustering.The authors propose a weighted KL difference as category choice function.The merging criterion is used as learning of category parameters.

Improved Gaussian Mixture CPHD Tracker for Multitarget Tracking

The Gaussian mixture cardinality probability hypothesis density (GM-CPHD) tracker is a promising algorithm for multitarget tracking. However, there are two major problems with it. First, when missed detections occur, the probability hypothesis density (PHD) weight will be shifted from the undetected part to the detected part, no matter how far apart the parts are. Second, when targets are close to or cross each other, the GM-CPHD tracker may fail to discriminate different tracks because the score of each track hypothesis in the traditional method is updated by simply summing the log likelihood ratios (LLR) between successive scans. To solve these problems an improved GM-CPHD tracker is proposed that minimizes the effect of the weight shifting and subsequent estimation errors by a dynamic reweighting scheme and improves the performance of track continuity by a dynamic track management scheme. Simulation results show that the improved GM-CPHD tracker is superior to the traditional methods in both the aspects of target state estimate and maintenance of track continuity so that this improved GM-CPHD tracker will have good application prospects.

Improved Iterated-corrector PHD with Gaussian mixture implementation

Many filter algorithms based on the probability hypothesis density (PHD) filter have been proposed to solve the multi-target tracking (MTT) problem. Most of them are applied to single-sensor case. As a simple and feasible multi-sensor filter algorithm, the Iterated-PHD filter is influenced by the order of the sensor updates and the probability of detection. In this paper, an improved algorithm with a modified update formula is proposed to deal with the above problems. In this algorithm, the original detection probability is divided into two parts: the improved miss-detection probability and the improved detection probability, which take the order of the sensor updates and the original detection probability of each sensor into consideration simultaneously. The effectiveness of the proposed algorithm is verified by the simulation results. The paper presents an improved Iterated-corrector PHD algorithm, IIC-GM-PHD.The original detection probability is replaced by the improved miss detection probability and the improved detection probability.The authors propose a modified update formula for multi-sensor PHD.

A box-particle implementation of standard PHD filter for extended target tracking

The paper presents a box-particle implementation of the standard PHD filter. The proposed ET-Box-PHD filter is derived by the cell likelihood function defined. The proposed filter is suitable to the strong clutter surveillance areas. This paper presents a box-particle implementation of the standard probability hypothesis density (PHD) filter for extended target tracking, called the extended target box-particle PHD (ET-Box-PHD) filter. The proposed filter can dynamically track multiple extended targets and estimate the unknown number of extended targets, in the presence of clutter measurements, false alarms and missed detections, where the extended targets are described as a Poisson model developed by Gilholm et al. To get the PHD recursion of the ET-Box-PHD filter, a suitable cell likelihood function for one given reliable partition is derived, and the main filter steps are presented along with the necessary box manipulations and approximations. The capabilities and limitations of the proposed ET-Box-PHD filter are illustrated both in linear simulation examples and in nonlinear ones. The simulation results show that the proposed ET-Box-PHD filter can effectively avoid the high number of particles and obviously reduce computational burden, compared to a particle implementation of the standard PHD filter for extended target tracking.

A global difference measure for the reduction of Gaussian inverse Wishart mixtures

This paper presents an evaluation criterion, called a global difference measure, for the reduction of Gaussian inverse Wishart (GIW) mixtures. It is a deviation between the original and reduced GIW mixture, in other words, a numerical way describing the performance of the reduction algorithm instead of just a previous curve analysis (i.e., visual inspection of the resulting mixture intensity functions). In this paper, the global difference measure is obtained by solving the normalized integrated squared distance (NISD). Additionally, a weighted Kullback-Leibler (KL) difference for the reduction of GIW is proposed, which makes a small modification to an existing algorithm introduced by Granstrom et al. (2012) [1]. The weighted KL difference is derived by considering the weights of components. This is ignored in the existing literatures. Both the proposed evaluation criterion and algorithm are tested on simulation examples, and the results show that the proposed evaluation criterion can depict correctly the result of the curve analysis.

Distributed multi-sensor particle filter for bearings-only tracking

In this article, the classical bearings-only tracking (BOT) problem for a single target is addressed, which belongs to the general class of non-linear filtering problems. Due to the fact that the radial distance observability of the target is poor, the algorithm-based sequential Monte-Carlo (particle filtering, PF) methods generally show instability and filter divergence. A new stable distributed multi-sensor PF method is proposed for BOT. The sensors process their measurements at their sites using a hierarchical PF approach, which transforms the BOT problem from Cartesian coordinate to the logarithmic polar coordinate and separates the observable components from the unobservable components of the target. In the fusion centre, the target state can be estimated by utilising the multi-sensor optimal information fusion rule. Furthermore, the computation of a theoretical Cramer–Rao lower bound is given for the multi-sensor BOT problem. Simulation results illustrate that the proposed tracking method can provide better performances than the traditional PF method.