Ioannis Kyriakides

Sequential Monte Carlo Methods for Tracking Multiple Targets With Deterministic and Stochastic Constraints

In multitarget scenarios, kinematic constraints from the interaction of targets with their environment or other targets can restrict target motion. Such motion constraint information could improve tracking performance if effectively used by the tracker. In this paper, we propose three particle filtering methods that incorporate constraint information in their proposal and weighting process; the number of targets is fixed and known in all methods. The reproposed constrained motion proposal (RCOMP) utilizes an accept/reject method to propose particles that meet the constraints. The truncated constraint motion proposal (TCOMP) uses proposal densities truncated to satisfy the constraints. The constraint likelihood independent partitions (CLIP) method simply rejects proposed partitions that do not meet the constraints. We use simulation to evaluate the performance of these three methods for two constrained motion scenarios: a vehicle convoy and soldiers executing a leapfrog motion. Moreover, we demonstrate the utility of constraint information by comparing the proposed algorithms with the independent partition (IP) proposal method that does not use constraint information. The simulation results demonstrate that the root mean square error (RMSE) tracking performance of the RCOMP and the TCOMP methods is much better than the CLIP and IP methods; this is due to their more efficient proposal process.

Monte Carlo Matching Pursuit Decomposition Method for Damage Quantification in Composite Structures

In wave-based approach, the presence of damage is visualized in terms of the changes in the signature of the resultant wave that propagates through the structure. In structural health monitoring, the fundamental goal is to detect, localize, and quantify these damage signatures. The current approach uses matching pursuit decomposition (MPD) to compare signals from healthy and damaged structures. However, the major drawback of the MPD is that, in the decomposition process, it performs an exhaustive search over a large dictionary of elementary functions. Therefore, this method of decomposition is associated with a large computational expense. In this research, the Monte Carlo matching pursuit decomposition (MCMPD) is proposed, that adapts a smaller dictionary to the signal structure, thus avoiding the exhaustive search over the time-frequency plane. The proposed algorithm, sequentially estimates a dictionary that contains only those components that match the waveform structure, uses the matching pursuits for the decomposition of the signal and if necessary, adapts the dictionary to the structure of the residues for further decomposition. Finally, we demonstrate using real life data that the MCMPD retains the ability of the matching pursuit to decompose waveforms and quantify them accurately while reducing computational expense.

Target tracking using particle filtering and CAZAC sequences

When tracking targets in radar, the selection of the transmitted waveform and the method of processing the return signal are two of the design aspects that affect measurement accuracy. Increased measurement accuracy results in enhanced tracking performance. In this paper, we apply sequential Monte Carlo methods to propose matched filtering operations in the delay-Doppler space where a target is expected to exist. Moreover, in the case of thresholding the measurements, these methods are used to form resolution cells that have the shape of the probability of detection contour. These methods offer an advantage over traditional radar tracking methods that form tessellating resolution cells to approximate the probability of detection contours, and exhaustively perform matched filtering operations over the entire delay-Doppler space. With the use of a Bjorck constant amplitude zero-autocorrelation (CAZAC) sequence, a high resolution measurement is attained and the use of thresholding is avoided. This is an advantage over commonly used waveforms such as linear frequency modulated chirps (LFMs). We examine the properties of Bjorck CAZACs and demonstrate improved tracking performance over LFMs in a single target tracking scenario.

A Particle Filtering Approach To Constrained Motion Estimation In Tracking Multiple Targets

Particle filtering has been successfully used in complex target tracking applications such as multiple target tracking. The particle filter can be used to incorporate constraints on target motion to improve tracking performance; this can be achieved using likelihood functions and sampling distributions. In this paper, we propose the constraint likelihood function independent partitions (CLIP) algorithm that uses constraints on target motion. This is achieved by incorporating a constraint likelihood function with the particle weights. As demonstrated by our simulations, a higher increase in tracking performance is obtained with our proposed constrained motion proposal (COMP) algorithm that incorporates target kinematic constraint information directly into the proposal density of the particle filter

Multiple target tracking with constrained motion using particle filtering methods

In this paper, we propose the constrained motion proposal (COMP) algorithm that incorporates target kinematic constraint information into a particle filter to track multiple targets. We represent deterministic or stochastic constraints on target motion as a likelihood function that is incorporated into the particle filter proposal density. Using Monte Carlo simulations, we demonstrate that this approach improves tracking performance while reducing computational cost relative to the independent partition particle filter with and without a constraint likelihood function.

Threshold Optimization for Distributed Detection using Particle Filtering Methods

Local processing on the nodes of a distributed sensing and processing system has the benefits of reducing the data volume transferred from the nodes to the fusion center, reducing both transmission power requirements and the computational burden on the fusion center. The individual nodes obtain measurements from the environment and transmit a quantized detection statistic to the fusion center. Quantization threshold levels need to be found for each sensor that maximize the performance of the system. We propose a global optimization method, the particle filtering optimization method, that uses particle filtering to propagate the values of the thresholds of a distributed detection system to sensor threshold values that are optimal with respect to some measure of system performance. We demonstrate, through simulations, the effectiveness of the particle filtering optimization method in finding the threshold of each of the sensors used in detection scenario

Adaptive High-Resolution Sensor Waveform Design for Tracking

Abstract Recent innovations in modern radar for designing transmitted waveforms, coupled with new algorithms for adaptively selecting the waveform parameters at each time step, have resulted in improvements in tracking performance. Of particular interest are waveforms that can be mathematically designed to have reduced ambiguity function sidelobes, as their use can lead to an increase in the target state estimation accuracy. Moreover, adaptively positioning the sidelobes can reveal weak target returns by reducing interference from stronger targets. The manuscript provides an overview of recent advances in the design of multicarrier phase-coded waveforms based on Bjorck constant-amplitude zero-autocorrelation (CAZAC) sequences for use in an adaptive waveform selection scheme for mutliple target tracking. The adaptive waveform design is formulated using sequential Monte Carlo techniques that need to be matched to the high resolution measurements. The work will be of interest to both practitioners and resear...

Using a Configurable Integrated Sensing And Processing Imager to Track Multiple Targets

On-line processing of data from video sequences is impeded by the need to process the large amount of data acquired by the image sensor. Appropriate processing onboard of the sensor can, however, reduce the data transmitted and processed by the tracker. In this paper, we use video sequences from an integrated sensing and processing (ISP) imager to track a variable number of targets; the imager is configured to select and filter data before transmission. The sensor configuration allows selection of (a) pixel blocks to be acquired and (b) filtering operations to be performed on the acquired blocks. We develop a configuration strategy using a particle filter to direct the acquisition and filtering. To track an unknown number of targets, we implement a multi-target tracker using a track-before-detect method. We demonstrate using simulations that the ISP approach reduces the data transmitted to the tracker without loss in tracking performance when compared to traditional sensing systems.

Adapting Matching Pursuit Dictionaries to Waveform Structure using Particle Filtering

Although the matching pursuit algorithm can accurately decompose waveforms, its use in real applications is limited. This is because it can be computationally intensive as it is based on selecting elements from complete dictionaries spanning the time-frequency plane of interest. There is, therefore, a need for smaller dictionaries that can still result in accurate waveform decompositions. In this paper, we propose the particle filter matching pursuit algorithm that adapts the dictionary to the waveform structure. This algorithm uses particle filtering, a sequential Monte Carlo approach, to estimate the dictionary suitable for the decomposition of a given waveform, and then uses the matching pursuit algorithm to decompose the waveform. We demonstrate, using simulations, that the particle filtering matching pursuit can decompose waveforms faster than the matching pursuit