We develop a framework to select a subset of sensors from a field in which the sensors have an ingrained independence structure. Given an arbitrary independence pattern, we construct a graph that denotes pairwise independence between sensors, which means those sensors may operate simultaneously. The set of all fully-connected subgraphs (cliques) of this independence graph forms the independent sets of a matroid over which we maximize the minimum of a set of submodular objective functions. We propose a novel algorithm called MatSat that exploits submodularity and, as a result, returns a near-optimal solution with approximation guarantees that are within a small factor of the average-case scenario. We apply this framework to ping sequence optimization for active multistatic sonar arrays by maximizing sensor coverage and derive lower bounds for minimum probability of detection for a fractional number of targets. In these ping sequence optimization simulations, MatSat exceeds the fractional lower bounds and reaches near-optimal performance.

This paper presents the results of using a likelihood-based clustering step before tracking on a multistatic sonar step. The likelihood-based clustering appropriately models the measurement noise and allows for the incorporation of features. The clustering step also allows for the rejection of clutter and fusion of the contact measurements within a cluster. After clustering, fusion and classification, the tracking results are improved over previous preprocessing methods. Results are shown for the three scenarios in the PACSim dataset.

Recently, researchers at the Applied Physics Laboratory at the University of Washington collected a unique dataset by suspending two cameras, one infrared and one electro-optical, from a balloon. This apparatus was then used to image objects drifting on the surface of Lake Washington. The authors took that data and built a processing stream to track the movements of those drifting surface objects.

Recently a data set was collected using an imaging sonar of a non-stationary underwater object. This paper presents the image processing algorithms as well as the tracking algorithms used to take the imaging sonar data and track a non-stationary underwater extended object. The tracking results will be presented in a geo-referenced image frame with the use of GPS and inertial sensors. Future work with this data set will include feature extraction and object classification using the imaging sonar data.

This paper presents approaches for incorporating classification information into target tracking algorithms, specifically in a multistatic active sonar context. In addition, this paper describes the framework designed for simulation and classification of return time series from simulated targets and clutter in a realistic underwater environment. The simulated target and clutter returns are integrated into an existing contact-based tracking dataset (TNO Blind dataset) for which time series are unavailable. Simulations compare the integrating classification of contacts at different stages of tracking algorithms. Results show improvements in some tracking metrics with no degradation of the others.

We implement and evaluate a method to infer position from Doppler measurements in a multistatic sonar scenario and present a likelihood approach for doing so. Doppler measurements are used to create likelihood surfaces for each of the transmitter–receiver pairs. The likelihood surfaces are combined and can then be used as-is or combined with additional position measurements. The final likelihood surface is usable in a Bayesian-style tracker or can be used to estimate position of a contact for use in a contact-based tracker. We show how the estimate improves with the addition of multiple receivers and show how the use of Doppler information can improve tracking results.

This paper presents tracking results on the Metron data set using the JPDA algorithm and a preprocessing likelihood surface formulation. The Metron data set is a simulated data set and is designed to be very difficult with large bearing and range errors which leads to high localization error for true detections. There are also significant amounts of clutter. Results using other data association algorithms such as the PDA, PDAFAI, and PDAFAIwTS were not good, which led to the use of a likelihood surface. The preprocessing step using the likelihood surface is key for achieving reasonable results. For the baseline tracking scenario where the truth is known, the results were encouraging. Extending this technique to include acoustic modeling and Doppler information will be topics of future research.

In this communication, the problem of determining effective pinging strategies in multistatic sonar systems with multiple transmitters is addressed. New algorithms are presented to determine effective pinging strategies for generalized search scenarios. An important part of this work is the development of metrics to be used in the optimization procedures. For maintaining search coverage, a ldquoprobability of target presencerdquo metric formulation is used. This formulation utilizes sonar performance prediction and a Bayesian update to incorporate negative information (i.e., searching an area but finding no targets) into the optimization procedure. The possibility of targets moving into previously searched areas is accounted for by using a Fokker-Planck (FP) drift/diffusion formulation. Monte Carlo simulations are used to show the accuracy and efficiency of this formulation. This formulation is shown to be computationally efficient compared to Monte Carlo simulations. It is also demonstrated that by choosing the ping sequence intelligently, the field performance can be improved compared to random or sequential ping sequencing.

Underwater targets in multistatic active sensor fields using the Probabilistic Data Association Filter with
Amplitude Information (PDAFAI). Target strength distributions as a function of aspect of expected targets are used to develop a new characterization of the signal-plus-noise probability density function that can account for knowledge of a target's aspect as a track develops or for the lack of such knowledge for newly created tracks. These density functions are then used in the amplitude likelihood ratio to provide
more accurate data association.

We propose a Bayesian estimation method to sequentially update the probability of detection for tracking. A beta distribution is used for the prior, which can be centered on the best a priori guess for the probability of detection. The tracker%u2019s belief about whether it detected the target at the last scan is used to update the posterior estimate of the probability of detection. The method can be applied to any tracking algorithm that requires an estimate of the probability of detection. Experiments with the probabilistic data association (PDA) tracker show that the proposed estimation method can increase the amount of time a target is tracked and decrease the localization error when compared to using a fixed value. Experiments also show that for some values of the probability of detection, using an inflated value of the probability of detection in PDA can actually lead to better performance.

This paper presents an approach for adapting a tracking algorithm to the acoustic propagation environment. This adaptation is performed by incorporating the expected target signal-to-noise ratio (SNR) into the data association step through the measured contact amplitude. In this work, expected SNR is provided via acoustic modeling; estimates of bottom loss and scattering strength, required by the acoustic model, are obtained via inversion of the acoustic model based on measured multi-static sonar reverberation data. This paper shows that the use of distributed sensors provides improved estimates of the environmental parameters, and hence better estimates of the expected SNR.

Preprocessing of data to be learned by a neural network is typically done to improve neural network performance. Output processing is especially important since it directly affects the influence of error in the hidden layers on the error of the neural network output. Principal component analysis is a commonly used preprocessing method that can improve the network performance by reducing the output dimensionality and reducing the number of parameters in a neural network model. Transforming the principal components of the outputs with an orthonormal matrix prior to scaling can further improve network performance.