Terry L. Henderson

Application of state-variable techniques to optimal feature extraction--- Multichannel analog data

In problems of pattern recognition and signal detection, one of the most important tasks is that of finding practical ways of preprocessing the data to eliminate complexity, redundancy, and irrelevancy. In this paper it is assumed that a vector wavefonn is received during an interval [t_o, t_f] . The waveform is considered to be a sample of a nonstationary vector random process containing a signal process and a noise process consisting of both white and colored noise. The optimum set of weighting functions is found for integrating the received waveform to extract those features that best reveal the presence of the signal. The solution is also shown to be the optimum one for estimating signal strength. A practical scheme for obtaining the optimum weighting functions is derived using state variables, and worked examples are presented.

Doppler estimation and correction for shallow underwater acoustic communications

Reliable mobile underwater acoustic communication systems must compensate for strong, time-varying Doppler effects. Many Doppler correction techniques rely on a single bulk correction to compensate first-order effects. In many cases, residual higher-order effects must be tracked and corrected using other methods. The contributions of this paper are evaluations of (1) signal-to-noise ratio (SNR) performance from three Doppler estimation and correction methods and (2) communication performance of Doppler correction with static vs. adaptive equalizers. The evaluations use our publicly available shallow water experimental dataset, which consists of 360 packet transmission samples (each 0.5s long) from a five-channel receiver array.

Comments on linear feature extraction [Corresp.]

The problem considered is that of finding the best linear transformation to reduce a random-data vector z to a vector of smaller dimension. It is assumed that the original data are Gaussian under either of two hypotheses, and that one wishes to use the transformed data to distinguish the hypotheses. The Bhattacharya distance is used to measure the information carried by the transformed data. A compromise solution is obtained for the case in which the data have both different means and different covariances under the alternative hypotheses.

Wideband monopulse spatial filtering for large receiver arrays for reverberant underwater communication channels

Underwater acoustic communication (ACOMMS) is critical for many applications including marine science, oceanographic exploration, offshore surveying/drilling, and military uses. ACOMMS data rates are usually limited by multiple propagation paths with different time delays and Doppler characteristics. It is often difficult to coherently recombine all paths, especially in shallow water, leaving incoherent paths that interfere with the receiver. One way to suppress unwanted paths is with a directional receiving array. Indeed, many existing large, directional acoustic arrays could be used as ACOMMS receivers. In a number of these arrays, wideband monopulse outputs could be made available. These directional beam outputs, in monopulse pairs, can selectively suppress, or even null, offending multipath when combined with a simple scalar weight. Using an experimental system, we show how a relatively short equalizer, using as inputs the wideband monopulse beam outputs of a large array, can form the backbone of an ACOMMS system that performs effectively in a multipath-limited environment. Our contributions include (i) a multipath-Doppler channel model validated by experimental results, (ii) a receiver design that utilizes monopulse processing, and (iii) an analysis of its performance using simulated and experimental data.

Rank Reduction For Broadband Waves Incident On A Linear Receiving Aperture

following form: Express a vector function a(t) (or a matrix A) as the product of a constant, rectangular Vandermonde matrix V and a vector function c(t) (or a matrix C). Transformations can be applied to a(t) (or A) to change V, potentially reducing its rank. In processing multichannel outputs from a linear receiving aperture, such a transformation can reduce the rank to the number of plane wave sources present, irrespective of the number of frequencies that each source emits.

Multi-stage and sparse equalizer design for communication systems in reverberant underwater channels

Coherent underwater communication systems in shallow water must compensate for several impairments, including Doppler shift and reverberant channels. In this paper, we quantify tradeoffs in communication performance vs. computational complexity in designing receivers to compensate for these impairments. Our communication system is unidirectional, single-carrier, wideband, and packet-based. We use 1.5 hours of recorded data from an experimental system on a public lake with a moving transmitter and stationary receiver. Our contributions include acoustic channel modeling and a tradeoff analysis for Doppler shift estimation, multi-stage equalizers and sparse equalizers. We compare multi-stage and sparse equalizers against traditional equalizers.

Digital matched filters for detecting Gaussian signals in Gaussian noise

The use of a set of digital matched filters is presented as an alternative to direct computation of the likelihood-ratio, for the problem of detecting a random signal in random noise. It is assumed that a random process composed of Gaussian background noise and (with probability P) a zero-mean Gaussian signal is sampled at N instants, the samples being corrupted by additive Gaussian measurement noise. The samples are processed by K @? N digital correlation filters which are structured so that the signal can be detected with minimum Bayes risk. The optimum filters are shown to be matched to the most relevant components of the simultaneously orthogonal expansion of the set of sampled data. State variable techniques are used to find a very practical method for determining the optimum filter structures.

A space–time filtered gradient method for detecting directions of echoes and transient soundsa)

The gradient vector (e.g., of the acoustic pressure) indicates the direction to the source of a wave, but it is easily corrupted by interference from other directions. However the gradient concept, even for higher orders, can be applied rigorously to a beamforming aperture that shields against interference, thereby allowing precise determination of the direction of sound echoes or emissions, especially for very brief, broadband transient sounds. In this treatment there is no gradient sensor per se; the aperture weighting supplants that function. Various geometric shapes can be used as apertures, but simple plates are often best, and the required weightings can be realized by patterned electrodes. The method is shown to be a natural extension of earlier techniques and inventions, and useful interpretations and generalizations are provided, such as compound and steered apertures, instantantly re-steerable nulls, and an equivalence to tracking acoustic particle motion after acoustical shielding from interfer...

Wideband direction finding via shielded gradient beamspace techniques

Monopulse techniques have been used for over 50 years in the radar community to estimate the direction of arrival (DOA) of incoming echoes. In recent years, a variant of the monopulse technique has been developed, termed the shielded gradient technique, which allows DOA estimation for signals of arbitrary bandwidth. The technique maps the array‐output M‐vector into a frequency‐invariant B‐dimensional beamspace. The work presented here describes the shielded gradient beamspace model in its higher‐order form, and develops wideband DOA estimation algorithms analogous to the narrow‐band MUSIC, root‐MUSIC, and ESPRIT algorithms. The performance of these new algorithms is studied through simulation and application to measured, in‐water sonar data. They are also compared via simulation to existing wideband DOA estimation algorithms. [Work supported by the Internal Research and Development Program under Contract No. FEE‐800.]

Time‐frequency analysis of low diffraction beam pulses

Recent work in low diffraction beams has given our group some new sets of wave fields to investigate. Theoretical predictions matched experimental measurements closely, and several comparison techniques were included in the analysis. The field experiments have now been conducted at the NUWC Keyport facility and, most recently, at the University of Texas’ Applied Research Laboratory facility in Lake Travis. In cooperation with ARL, a series of low diffraction beam pulses was launched to a wide band width receiver 600 feet distant. Among the comparison methods used was a time‐frequency analysis of the received pulse. For each angular position the frequency arrival times via time‐frequency analysis are investigated. It is hoped this analysis will shed light on this very interesting phenomena.