The global optimization of phase-incoherent signals

The problem of global optimization of M phase-incoherent signals in N complex dimensions is formulated. Then, by using the geometric approach of Landau and Slepian, conditions for optimality are established for N = 2 and the optimal signal sets are determined for M = 2, 3, 4, 6, and 12.

The method is the following: The signals are assumed to be equally probable and to have equal energy, and thus are represented by points ṡ_i, i = 1, 2, …, M, on the unit sphere S₁ in C^N. If W_ik is the halfspace determined by ṡ_i and ṡ_k and containing ṡ_i, i.e. W_ik = {ṙϵC^N:| ≥ | ˂ṙ, ṡ_k˃|}, then the Ʀ_i = ∩/k≠i W_ik, i = 1, 2, …, M, the maximum likelihood decision regions, partition S₁. For additive complex Gaussian noise ṅ and a received signal ṙ = ṡ_ie^jϴ + ṅ, where ϴ is uniformly distributed over [0, 2π], the probability of correct decoding is P_C = 1/π^N ∞/ʃ/0 r^2N-1e^-(r2+1)U(r)dr, where U(r) = 1/M M/Ʃ/i=1 Ʀ_i ʃ/∩ S₁ I₀(2r | ˂ṡ, ṡ_i˃|)dσ(ṡ), and r = ǁṙǁ.

For N = 2, it is proved that U(r) ≤ ʃ/C_α I₀(2r|˂ṡ, ṡ_i˃|)dσ(ṡ) – 2K/M. h(1/2K [Mσ(C_α)-σ(S₁)]), where C_α = {ṡϵS₁:|˂ṡ, ṡ_i˃| ≥ α}, K is the total number of boundaries of the net on S₁ determined by the decision regions, and h is the strictly increasing strictly convex function of σ(C_α∩W), (where W is a halfspace not containing ṡ_i), given by h = ʃ/C_α∩W I₀ (2r|˂ṡ, ṡ_i˃|)dσ(ṡ). Conditions for equality are established and these give rise to the globally optimal signal sets for M = 2, 3, 4, 6, and 12.

Bayesian approach to signal modelling

In this paper, an introduction to Bayesian methods in signal processing will be given. The paper starts by considering the important issues of model selection and parameter estimation and derives analytic expressions for the model probabilities of two simple models. The idea of marginal estimation of certain model parameter is then introduced and expressions are derived for the marginal probability densities for frequencies in white Gaussian noise and a Bayesian approach to general changepoint analysis is given. Numerical integration methods are introduced based on Markov chain Monte Carlo techniques and the Gibbs sampler in particular.

The Bayesian Approach to Signal Modelling

In this paper, an introduction to Bayesian methods in signal processing will be given. The paper starts by considering the important issues of model selection and parameter estimation and derives analytic expressions for the model probabilities of two simple models. The idea of marginal estimation of certain model parameter is then introduced and expressions are derived for the marginal probabilitiy densities for frequencies in white Gaussian noise and a Bayesian approach to general changepoint analysis is given. Numerical integration methods are introduced based on Markov chain Monte Carlo techniques and the Gibbs sampler in particular.

Delayed-decision making in noisy bistable switches

Bistable switches are frequently encountered in biological systems. Typically, a bistable switch models a binary decision where each decision corresponds to one of the two stable equilibria. Recently, we showed that the global decision-making process in bistable switches strongly depends on a particular equilibrium point of these systems, their saddle point. In particular, we showed that a saddle point with a time-scale separation between its attractive and repulsive directions can delay the decision-making process. In this paper, we study the effects of white Gaussian noise on this mechanism of delayed decision-making induced by the saddle point. Results show that the mean decision-time strongly depends on the balance between the initial distance to the separatrix and the noise strength. © IFAC.

Classification of lactose and mandelic acid THz spectra using subspace and wavelet-packet algorithms

This work compares classification results of lactose, mandelic acid and dl-mandelic acid, obtained on the basis of their respective THz transients. The performance of three different pre-processing algorithms applied to the time-domain signatures obtained using a THz-transient spectrometer are contrasted by evaluating the classifier performance. A range of amplitudes of zero-mean white Gaussian noise are used to artificially degrade the signal-to-noise ratio of the time-domain signatures to generate the data sets that are presented to the classifier for both learning and validation purposes. This gradual degradation of interferograms by increasing the noise level is equivalent to performing measurements assuming a reduced integration time. Three signal processing algorithms were adopted for the evaluation of the complex insertion loss function of the samples under study; a) standard evaluation by ratioing the sample with the background spectra, b) a subspace identification algorithm and c) a novel wavelet-packet identification procedure. Within class and between class dispersion metrics are adopted for the three data sets. A discrimination metric evaluates how well the three classes can be distinguished within the frequency range 0. 1 - 1.0 THz using the above algorithms.

Observations on the propagation of internal tidal waves on the continental slope off Sierra Leone

During R/V Meteor-cruise no. 30 4 moorings with 17 current meters were placed on the continental slope of Sierra Leone at depths between 81 and 1058 meters. The observation period started on March 8, 1973, 16.55 hours GMT and lasted 19 days for moorings M30_068MOOR, M30_069MOOR, M30_070MOOR on the slope and 9 days for M30_067MOOR on the shelf. One current meter recorded at location M30_067MOOR for 22 days. Hydrographic data were collected at 32 stations by means of the "Kieler Multi-Meeressonde". Harmonic analysis is applied to the first 15 days of the time series to determine the M2 and S2 tides. By vertically averaging of the Fourier coefficients the field of motion is separated into its barotropic and its baroclinic component. The expected error generated by white Gaussian noise is estimated. To estimate the influence of the particular vertical distribution of the current meters, the barotropic M2 tide is calculated by ommitting and interchanging time series of different moorings. It is shown that only the data of moorings M30_069MOOR, M30_070MOOR and M30_067MOOR can be used. The results for the barotropic M2 tide agree well with the previous publications of other authors. On the slope at a depth of 1000 m there is a free barotropic wave under the influence of the Coriolis-force propagating along the slope with an amplitude of 3.4 cm S**-1. On the shelf, the maximum current is substantially greater (5.8 cm s**-1) and the direction of propagation is perpendicular to the slope. As for the continental slope a separation into different baroclinic modes using vertical eigenmodes is not reasonable, an interpretation of the total baroclinic wave field is tried by means of the method of characteristis. Assuming the continental slope to generate several linear waves, which superpose, baroclinic tidal ellipses are calculated. The scattering of the direction of the major axes M30_069MOOR is in contrast to M30_070MOOR, where they are bundled within an angle of 60°. This is presumably caused by the different character of the bottom topography in the vicinity of the two moorings. A detailed discussion of M30_069MOOR is renounced since the accuracy of the bathymetric chart is not sufficient to prove any relation between waves and topography. The bundeling of the major axes at M30_070MOOR can be explained by the longslope changes of the slope, which cause an energy transfer from the longslope barotropic component to the downslope baroclinic component. The maximum amplitude is found at a depth of 245 m where it is expected from the characteristics originating at the shelf edge. Because of the dominating barotropic tide high coherence is found between most of the current meters. To show the influence of the baroclinic tidal waves, the effect of the mean current is considered. There are two periods nearly opposite longshore mean current. For 128 hours during each of these periods, starting on March 11, 05.00, and March 21, 08.30, the coherences and energy spectra are calculated. The changes in the slope of the characteristics are found in agreement with the changes of energy and coherence. Because of the short periods of nearly constant mean current, some of the calculated differences of energy and coherence are not statistically significant. For the M2 tide a calculation of the ratios of vertically integrated total baroclinic energy and vertically integrated barotropic kinetic energy is carried out. Taking into account both components (along and perpendicular to the slope) the obtained values are 0.75 and 0.98 at the slope and 0.38 at the shelf. If each component is considered separately, the ratios are 0.39 and 1.16 parallel to the slope and 5.1 and 15.85 for the component perpendicular to it. Taking the energy transfer from the longslope component to the doenslope component into account, a simple model yields an energy-ratio of 2.6. Considering the limited application of the theory to the real conditions, the obtained are in agreement with the values calculated by Sandstroem.

Novel method to improve the signal-to-noise ratio in the far-field results obtained from planar near-field measurements

A method to reduce the noise power in far-field pattern without modifying the desired signal is proposed. Therefore, an important signal-to-noise ratio improvement may be achieved. The method is used when the antenna measurement is performed in planar near-field, where the recorded data are assumed to be corrupted with white Gaussian and space-stationary noise, because of the receiver additive noise. Back-propagating the measured field from the scan plane to the antenna under test (AUT) plane, the noise remains white Gaussian and space-stationary, whereas the desired field is theoretically concentrated in the aperture antenna. Thanks to this fact, a spatial filtering may be applied, cancelling the field which is located out of the AUT dimensions and which is only composed by noise. Next, a planar field to far-field transformation is carried out, achieving a great improvement compared to the pattern obtained directly from the measurement. To verify the effectiveness of the method, two examples will be presented using both simulated and measured near-field data.

Direction of arrival estimation in the presence of distributed noise sources: Cumulant based approach

In the past few years there have been attempts to develop subspace methods for DoA (direction of arrival) estimation using a fourth?order cumulant which is known to de?emphasize Gaussian background noise. To gauge the relative performance of the cumulant MUSIC (MUltiple SIgnal Classification) (c?MUSIC) and the standard MUSIC, based on the covariance function, an extensive numerical study has been carried out, where a narrow?band signal source has been considered and Gaussian noise sources, which produce a spatially correlated background noise, have been distributed. These simulations indicate that, even though the cumulant approach is capable of de?emphasizing the Gaussian noise, both bias and variance of the DoA estimates are higher than those for MUSIC. To achieve comparable results the cumulant approach requires much larger data, three to ten times that for MUSIC, depending upon the number of sources and how close they are. This is attributed to the fact that in the estimation of the cumulant, an average of a product of four random variables is needed to make an evaluation. Therefore, compared to those in the evaluation of the covariance function, there are more cross terms which do not go to zero unless the data length is very large. It is felt that these cross terms contribute to the large bias and variance observed in c?MUSIC. However, the ability to de?emphasize Gaussian noise, white or colored, is of great significance since the standard MUSIC fails when there is colored background noise. Through simulation it is shown that c?MUSIC does yield good results, but only at the cost of more data.

Model-based approaches to handling additive noise in reverberant environments

Model-based approaches to handle additive and convolutional noise have been extensively investigated and used. However, the application of these schemes to handling reverberant noise has received less attention. This paper examines the extension of two standard additive/convolutional noise approaches to handling reverberant noise. The first is an extension of vector Taylor series (VTS) compensation, reverberant VTS, where a mismatch function including reverberant noise is used. The second scheme modifies constrained MLLR to allow a wide-span of frames to be taken into account and projected into the required dimensionality. To allow additive noise to be handled, both these schemes are combined with standard VTS. The approaches are evaluated and compared on two tasks, MC-WSJ-AV, and a reverberant simulated version of AURORA-4. © 2011 IEEE.

Gaussian Process training with input noise

In standard Gaussian Process regression input locations are assumed to be noise free. We present a simple yet effective GP model for training on input points corrupted by i.i.d. Gaussian noise. To make computations tractable we use a local linear expansion about each input point. This allows the input noise to be recast as output noise proportional to the squared gradient of the GP posterior mean. The input noise variances are inferred from the data as extra hyperparameters. They are trained alongside other hyperparameters by the usual method of maximisation of the marginal likelihood. Training uses an iterative scheme, which alternates between optimising the hyperparameters and calculating the posterior gradient. Analytic predictive moments can then be found for Gaussian distributed test points. We compare our model to others over a range of different regression problems and show that it improves over current methods.

A novel shrinkage technique based on the normal inverse Gaussian density model

This paper proposes a novel image denoising technique based on the normal inverse Gaussian (NIG) density model using an extended non-negative sparse coding (NNSC) algorithm proposed by us. This algorithm can converge to feature basis vectors, which behave in the locality and orientation in spatial and frequency domain. Here, we demonstrate that the NIG density provides a very good fitness to the non-negative sparse data. In the denoising process, by exploiting a NIG-based maximum a posteriori estimator (MAP) of an image corrupted by additive Gaussian noise, the noise can be reduced successfully. This shrinkage technique, also referred to as the NNSC shrinkage technique, is self-adaptive to the statistical properties of image data. This denoising method is evaluated by values of the normalized signal to noise rate (SNR). Experimental results show that the NNSC shrinkage approach is indeed efficient and effective in denoising. Otherwise, we also compare the effectiveness of the NNSC shrinkage method with methods of standard sparse coding shrinkage, wavelet-based shrinkage and the Wiener filter. The simulation results show that our method outperforms the three kinds of denoising approaches mentioned above.

Dynamics of Gompertzian tumour growth under environmental fluctuations

We investigate the effect of correlated additive and multiplicative Gaussian white noise oil the Gompertzian growth of tumours. Our results are obtained by Solving numerically the time-dependent Fokker-Planck equation (FPE) associated with the stochastic dynamics. In Our numerical approach we have adopted B-spline functions as a truncated basis to expand the approximated eigenfunctions. The eigenfunctions and eigenvalues obtained using this method are used to derive approximate solutions of the dynamics under Study. We perform simulations to analyze various aspects, of the probability distribution. of the tumour cell populations in the transient- and steady-state regimes. More precisely, we are concerned mainly with the behaviour of the relaxation time (tau) to the steady-state distribution as a function of (i) of the correlation strength (lambda) between the additive noise and the multiplicative noise and (ii) as a function of the multiplicative noise intensity (D) and additive noise intensity (alpha). It is observed that both the correlation strength and the intensities of additive and multiplicative noise, affect the relaxation time.

Methods for noise reduction in far-field patterns obtained from cylindrical near-field antenna measurements

Two different methods to reduce the noise power in the far-field pattern of an antenna as measured in cylindrical near-field (CNF) are proposed. Both methods are based on the same principle: the data recorded in the CNF measurement, assumed to be corrupted by white Gaussian and space-stationary noise, are transformed into a new domain where it is possible to filter out a portion of noise. Those filtered data are then used to calculate a far-field pattern with less noise power than that one obtained from the measured data without applying any filtering. Statistical analyses are carried out to deduce the expressions of the signal-to-noise ratio improvement achieved with each method. Although the idea of the two alternatives is the same, there are important differences between them. The first one applies a modal filtering, requires an oversampling and improves the far-field pattern in all directions. The second method employs a spatial filtering on the antenna plane, does not require oversampling and the far-field pattern is only improved in the forward hemisphere. Several examples are presented using both simulated and measured near-field data to verify the effectiveness of the methods.

Regression with input-dependent noise: A Bayesian treatment

In most treatments of the regression problem it is assumed that the distribution of target data can be described by a deterministic function of the inputs, together with additive Gaussian noise having constant variance. The use of maximum likelihood to train such models then corresponds to the minimization of a sum-of-squares error function. In many applications a more realistic model would allow the noise variance itself to depend on the input variables. However, the use of maximum likelihood to train such models would give highly biased results. In this paper we show how a Bayesian treatment can allow for an input-dependent variance while overcoming the bias of maximum likelihood.

Statistics of noise-driven coupled nonlinear oscillators:applications to systems with Kerr nonlinearity

We present exact analytical results for the statistics of nonlinear coupled oscillators under the influence of additive white noise. We suggest a perturbative approach for analysing the statistics of such systems under the action of a deterministic perturbation, based on the exact expressions for probability density functions for noise-driven oscillators. Using our perturbation technique we show that our results can be applied to studying the optical signal propagation in noisy fibres at (nearly) zero dispersion as well as to weakly nonlinear lattice models with additive noise. The approach proposed can account for a wide spectrum of physically meaningful perturbations and is applicable to the case of large noise strength. © 2005 Elsevier B.V. All rights reserved.