Donald E. Gustafson

A deeply integrated adaptive GPS-based navigator with extended range code tracking

This paper presents a new approach to GPS-based navigation which offers significant improvement in antijam capability over traditional designs. The algorithms may be implemented at low cost in software in existing and future GPS receivers using, as inputs, postcorrelation I and Q data and, optionally, raw data from other sensors. Traditional systems are not optimal at high jammer-to-signal (J/S) ratios as a consequence of modular design, use of traditional fixed-gain or gain-scheduled tracking loops, and use of artificial moding logic. The approach described here employs a nonlinear filter that operates efficiently at all J/S levels. Filter gains continuously adapt to changes in the J/S environment, and the error covariance propagation is driven directly by measurements to enhance robustness under high jamming and dynamics conditions. Extended-range correlation may be optionally included to increase the code tracking loss-of-lock threshold under high jamming scenarios. Computational complexity is comparable to an extended Kalman filter. Results of hardware-in-the-loop simulations are presented which demonstrate improvements of 15 dB or more in antijam capability relative to traditional designs.

ECG/VCG Rhythm Diagnosis Using Statistical Signal Analysis-I. Identification of Persistent Rhythms

In this paper we describe a technique for detection and classification of cardiac arrhythmias for ECG or VCG data. The approach is based on the use of R-R interval data and the development of simple models that describe the sequential behavior of such intervals characteristic of different arrhythmias which persist over a period of about six or more heartbeats. In the second part of this two-paper series, we will deal with arrhythmias that manifest themselves as abrupt changes in the observed R-R intervals.

Linear minimum variance filters applied to carrier tracking

A new approach is taken to the problem of tracking a fixed amplitude signal with a Brownian motion phase process. Classically, a first-order phase-lock loop (PLL) is used; here, the problem is treated via estimation of the quadrature signal components. In this space, the state dynamics are linear with white multiplicative noise. Therefore, linear, minimum-variance filters, which have a particularly simple mechanization, are suggested. The resulting error dynamics are linear at any signal/noise ( S/N ) ratio unlike the classical PLL. During synchronization, and above threshold, this filter with constant gains degrades by 3 percent in output rms phase error with respect to the classical loop. However, up to 80 percent of the maximum possible noise improvement is obtained below threshold where the classical loop is nonoptimum, as demonstrated by a Monte Carlo analysis. Filter mechanizations are presented for beth carrier and baseband operation. An interesting bandpass filter interpretation is given.

Innovative Indoor Geolocation Using RF Multipath Diversity

A new concept is presented for indoor geolocation in multipath environments where direct paths are sometimes undetectable. In contrast to previous statistically-based approaches, the multipath delays are modeled using a geometry-based argument. Assuming a series of specular reflections off of planar surfaces, the model contains a maximum of three unknown multipath parameters per path which may be estimated when geolocation accuracy is sufficiently high. If some of the direct paths subsequently become undetectable, it is possible under certain conditions to maintain geolocation accuracy using only the indirect path length measurements. The new concept is illustrated via simulation using a relatively simple representative scenario. Performance is compared to a traditional method which uses only direct path measurements, indicating the potential for significantly improved indoor geolocation accuracy in environments dominated by multipath. Since the estimated multipath parameters are geometry-dependent, this approach allows the possibility of building up indoor map information as the geolocation process commences.

A Nonlinear Code Tracking Filter for GPS-Based Navigation

Current Global Positioning System (GPS)-based navigation systems are highly susceptible to unintentional and intentional jamming due to relatively low signal power at the receiver antenna and, in part, due to suboptimal code tracking loop designs that do not account for measurement nonlinearities near loss-of-lock. A nonlinear code tracking filter is developed whose architecture is based on a rigorous minimum-variance solution of the navigation problem, rather than using prespecified tracking loop architectures. The filter implementation can be viewed in terms of the classical notions of error detector functions, which depend on signal-to-noise ratio (SNR) and root mean square (rms) code tracking error. Detector functions are defined for both code tracking error and code tracking error variance. The filter responds more rapidly than current designs to rapidly varying jammer power due to a measurement-dependent term in the covariance calculations. Extended-range tracking is utilized, yielding linear state vector error detector functions (i.e., the filter is essentially optimum) out to the maximum allowed by the correlator range, and reducing the need for reacquisition. Significant antijam improvements relative to current designs are predicted from high-fidelity simulation and hardware demonstrations. Computational requirements are comparable to extended Kalman filter/vector tracking loop techniques.

Robust incremental SLAM with consistency-checking

Incorrect landmark and loop closure measurements can cause standard SLAM algorithms to fail catastrophically. Recently, several SLAM algorithms have been proposed that are robust to loop closure errors, but it is shown in this paper that they cannot provide robust solutions when landmark measurement errors occur. The root cause of this problem is that the robust SLAM algorithms only focus on generating solutions that are locally consistent (i.e. each measurement agrees with its corresponding estimates) rather than globally consistent (i.e. all of the measurements in the solution agree with each other). Moreover, these algorithms do not attempt to maximize the number of correct measurements included in the solution, meaning that often correct measurements are ignored and the solution quality suffers as a result. This paper proposes a new formulation of the robust SLAM problem that seeks a globally consistent map that also maximizes the number of measurements included in the solution. In addition, a novel incremental SLAM algorithm, called incremental SLAM with consistency-checking, is developed to solve the new robust SLAM problem. Finally, simulated and experimental results show that the new algorithm significantly outperforms state-of-the-art robust SLAM methods for datasets with incorrect landmark measurements and can match their performance for datasets with incorrect loop closures.

A statistical approach to rhythm diagnosis of cardiograms

A new method is presented for detection and classification of arrhythmias in electrocardiograms on the basis of R-R interval data. A set of phenomenological models for both persistent and transient rhythms is developed to match observed statistical variations. Arrhythmias are identified by calculating statistical probabilities and likelihoods associated with these models using two recently developed techniques. The important system design considerations are described. Finally, representative results using actual arrhythmia data are presented to illustrate the system performance.

Indoor geolocation using RF multipath with probabilistic data association

An indoor navigation problem is considered where the objective is real-time tracking of transponding tags in multipath environments using signals sent from and received at a set of radio frequency (RF) sources at fixed and known locations. Current systems depend on detection of direct path signals and treat multipath signals as spurious. We present an approach for exploiting the multipath signals to maintain tracking when direct paths are undetected. The method relies on estimating the parameters of minimum-complexity models of the indirect path lengths. A maximum of three parameters is required to model indirect path lengths arising from an arbitrary number of specular reflections off planar surfaces. A probabilistic data association filter (PDAF) is used to mitigate uncertainties arising from noise, closely-spaced path lengths, and path length crossovers. The method is tested via simulation using bandlimited signals synthesized from ray trace data. Performance is compared to an optimized direct path filter using Monte Carlo analysis. No prior knowledge of multipath parameters or indoor infrastructure is assumed, and measurements are restricted to time-of-arrival (TOA) only. The results indicate that the PDAF consistently outperforms the direct path filter when one or more direct paths are blocked.

Linear transformations relating the electrocardiogram and vectorcardiogram

Abstract The average heartbeat waveforms of an individual, as obtained from the electrocardiogram (12 signals) and the vectorcardiogram (3 signals), are accurately related by a linear transformation. Depending on the individual, phase shifts between the electrocardiographic signals may have to be accounted for in the transformation; with this provision, excellent estimation of the average VCG heartbeat from the ECG record may be achieved. Attempts to find a single transformation which is valid for an entire group of individuals, however, have not been successful.

Feature selection for diagnosis of vectorcardiograms

The automatic classification of vectorcardiograms and electrocardiograms into disease classes using computerized pattern recognition techniques has been a much studied problem. To date, however, no system exists which meets desired accuracy and noise immunity requirements and development of new techniques continues. An important aspect of the problem is that of feature selection, in which the functions of data reduction and information preservation are performed. In this paper, the problem of linear feature extraction is studied and a modified form of the Karhunen-Loeve expansion is developed which appears to have some advantages for the present application. Comparison with other feature selection methods is made using a two-dimensional example. Finally, some areas for future research are pointed out.