Ilkka Karasalo

Estimating the covariance matrix by signal subspace averaging

An efficient algorithm is presented for estimating a covariance matrix consisting of a low-rank signal term and a full-rank noise term, known apart from a scalar factor. For each sample of the vector of sensor outputs, the algorithm approximates, in the least-squares sense, a rank-one update of the covariance matrix, under the side condition that the rank of the signal term remains bounded. If the model noise is spatially colored, the least-squares approximation is preceded by spatial prewhitening, It is shown that if the rank of the signal term is small compared to the number of sensors, then the proposed algorithm requires substantially less computational work than conventional averaging. Some simulation results are included, indicating that the proposed algorithm reduces the variance of some commonly used spectral estimators in off-target directions, without impairing their detection and resolution properties.

A criterion for truncation of theQR-decomposition algorithm for the singular linear least squares problem

A possible improvement of the Faddeev-Kublanovskaja-Faddeeva lower bound for the least singular value ofR by using additional information aboutR is discussed. A fast algorithm is given for calculating such a bound using the diagonal elements and the elements of largest modulus in each row ofR.

A high‐order adaptive integration method for wave propagation in range‐independent fluid–solid media

Efficient computation of the Hankel‐transform integral for the wave field in a laterally homogeneous fluid–solid medium is nontrivial, since the integrand may be both oscillating and irregularly peaked. We propose a high‐order, adaptive integration method suitable for integrands with these characteristics. The method combines trapezoidal or Filon sums, obtained with several step sizes, with polynomial or Bulirsch–Stoer rational extrapolation to increase the order of convergence and to obtain error estimates. This technique is combined with adaptive interval halving, maintaining a hierarchy of subintervals, meshes, and function values in a stack to eliminate duplicate function evaluations. Computational results from an underwater acoustics application are presented. At any level of accuracy, the proposed method requires less computational work than nonadaptive trapezoidal or Filon quadrature, the difference growing to orders of magnitude as the accuracy increases.

Long range sound propagation over a sea surface

This paper describes methodology and results from a model-based analysis of data on sound transmission from controlled sound sources at sea to a 10-km distant shore. The data consist of registrations of sound transmission loss together with concurrently collected atmospheric data at the source and receiver locations. The purpose of the analysis is to assess the accuracy of methods for transmission loss prediction in which detailed data on the local geography and atmospheric conditions are used for computation of the sound field. The results indicate that such sound propagation predictions are accurate and reproduce observed variations in the sound level as function of time in a realistic way. The results further illustrate that the atmospheric model must include a description of turbulence effects to ensure predicted noise levels to remain realistically high during periods of sound shadow.

On computing bounds for the least singular value of a triangular matrix

An algorithm for rapid computation of a lower bound for the least singular value of a triangular matrix is presented. It requiresO(N) operations whereN is the order of the matrix, and is based on the Perron-Frobenius theory of non-negative matrices. The input data are the diagonal elements and the off-diagonal elements of maximum modulus in each row.

Numerical study of a flame in a Kármán vortex street

Abstract A numerical study of a simplified flow configuration modeling the interaction of a premixed laminar flame with a Karman vortex street is presented. The flow is assumed to be two dimensional, inviscid everywhere except inside the vortices, and incompressible everywhere except at the flame. The flame is modeled as a single layer source for the velocity potential with a burning speed which depends on the local curvature. Complete flow fields are calculated as a function of time for various values of the Markstein parameter, vortex shedding frequency, separation of the vortex street, the strength of the vortices, and the ratio of unburned to burned fluid densities, ϱ u ϱ b . The velocity far upstream was set at 60 cm/s for all cases. The essential features of the interaction of a flame with a vortex street are demonstrated, including the deflection of the vortex street and streamlines ahead of the flame and the development of a flame brush. The results of the calculations show that predictions of the flame behavior are strongly dependent on the value of ϱ u ϱ b . It was also found that the frequency of the vortex street, the phase relations as the vortices pass through the flame, and the strength of the vortices are important parameters in determining the behavior of the flame.

Atmospheric Sound Propagation Over Large-Scale Irregular Terrain

A benchmark problem on atmospheric sound propagation over irregular terrain has been solved using a stable fourth-order accurate finite difference approximation of a high-fidelity acoustic model. A comparison with the parabolic equation method and ray tracing methods is made. The results show that ray tracing methods can potentially be unreliable in the presence of irregular terrain.

Long term estimations of low frequency noise levels over water from an off-shore wind farm.

This article focuses on computations of low frequency sound propagation from an off-shore wind farm. Two different methods for sound propagation calculations are combined with meteorological data for every 3 hours in the year 2010 to examine the varying noise levels at a reception point at 13 km distance. It is shown that sound propagation conditions play a vital role in the noise impact from the off-shore wind farm and ordinary assessment methods can become inaccurate at longer propagation distances over water. Therefore, this paper suggests that methodologies to calculate noise immission with realistic sound speed profiles need to be combined with meteorological data over extended time periods to evaluate the impact of low frequency noise from modern off-shore wind farms.

Transient Bistatic Scattering from Buried Objects

Computational predictions of bistatic scattering by spherical objects buried in sand underlying a water column are presented. Excitations with a high-frequency directive transient source both above and below critical grazing angle are considered. The predictions are compared with results from a recent tank experiment for investigating the field scattered by buried spherical objects. An ultrasonic Ricker-like waveform with center frequency 500 kHz was used and both solid and hollow spheres were studied.

A high-resolution postbeamforming method based on semiinfinite linear optimization

A method for enhancing the bearing resolution obtained by conventional beamforming is described. Using the beamformer power outputs, the beam response functions, and optionally a background-noise spatial power spectral density (PSD) as input, the method gives a point-mass estimate of the signal PSD as function of bearing. The estimate is defined from the solution of a semiinfinite linear optimization problem and can be computed efficiently by solving a sequence of linear programs obtained by discretization using adaptive mesh refinement. In theory the estimate can achieve arbitrary resolution of the bearings of distinct plane-wave signals in stationary noise. For actual hydroacoustic data, it has been shown to yield a significant improvement of the resolution obtained by traditional beamforming. Some results obtained with simulated data and with actual data from a hydroacoustic application are presented. >