Mittuniversitetet

Direction-of-arrival (DoA) estimation plays a critical role in applications of acoustic localization. This work proposes a closed-form algorithm for single-source one-dimensional direction of arrival estimation using the uniform circular array. The algorithm estimates the azimuthal direction of arrival using the least-square estimate of the circular harmonics steering vector. The least-square (LS) estimate of the steering vector is estimated as the principal eigenvector of the received data covariance matrix. The proposed algorithm is tested and compare against the time-frequency circular harmonics beamforming (TF-CHB) and eigenbeam MUSIC (EB-MUSIC) algorithms. The comparison shows that the proposed algorithm outperformed the two benchmark methods in accuracy under various test conditions. 

In this study, a hybridised bi-objective optimisation technique was proposed to optimise process parameters (additives) in pavement engineering. The two process parameters, rice husk ash (RHA) and quarry dust (QD), which were used for the treatment of the soft subgrade soil, were set as input parameters in the development of two regression functions with the RSM optimisation technique. Next, the developed regression functions were utilised as fitness functions in the MOGOA and Pareto optimal solutions that represent different optimum combinations of additives. Subsequently, a selected optimum combination of additives (14.5% RHA and 14.4% QD) was used to validate the proposed hybridised RSM-MOGOA technique. The predicted values by the RSM-MOGOA technique for the 28-day unconfined compressive strength (UCS) and the California bearing ratio (CBR) of the soft subgrade were 729.9 kN/m2 and 50.4%, respectively while those obtained from confirmatory experiment were 736.50 kN/m2 and 54.4%, respectively for the UCS and CBR. 

Heart sound classification is a critical task in automated cardiac diagnostics, yet it is often challenged by the limited availability of labeled data and the dominance of low-frequency components in heart sound signals. This study introduces a novel data augmentation technique, the first-difference method, to address these challenges in convolutional neural network (CNN)- based classification. By enhancing high-frequency components in the time domain, this method enables the model to better capture abnormalities, such as murmurs, present in higher frequency ranges. The effectiveness of this approach was evaluated using three spectral transformations—linear spectrogram, mel-spectrogram, and mel-frequency cepstrum coefficient (MFCC) —across multiple augmentation levels. The results demonstrate substantial improvements in classification metrics, including precision, recall, F1 score, and specificity, with MFCC-based predictors achieving the highest performance gains. The findings highlight the potential of the first-difference augmentation as a simple and effective strategy for improving heart sound classification, paving the way for more robust and generalizable diagnostic tools in real-world clinical applications.

Direction of arrival estimation using the spherical microphone array usually requires a search in the whole 3-dimensional space, hence computationally demanding. This work presents a machine learning approach to sectorizing the 3-dimensional space, as an intermediate step for direction-of-arrival estimation using spherical microphone array. A new feature based on the outer product of spherical harmonic vectors was proposed for the classification. This spherical harmonic matrix nominally offers lower dimensionality compared to the commonly used covariance matrix of received data. The dimension of the input matrix was further reduced using the neighborhood component analysis. The extracted features were then used to train a support vector machine (SVM), 2-layer multilayer perceptron (MLP) and a convolutional neural network (CNN) for classification purposes. The results show that the models were able to classify the spherical sector with up to 90 % accuracy for all models and number of sectors under consideration. Also, the MLP and CNN trained with simulated samples were able to accurately classify samples from real data that were not included in training samples.

The field of Metrology has seen great use of Machine Learning and Deep Learning models, improving existing Metrology and enabling measurements and estimations that were previously not possible. In the challenging task of gathering training data in various areas of Metrology, a question arises; is it necessary to gather a completely new dataset every time a quality upgrade is done to a measurement system, method or model, or can the formerly trained model be used for new data with higher Signal-to-Noise Ratio (SNR)? This paper investigates how trained neural networks react to new data coming into the testing, with a higher SNR than the training data. In the experiments, Convolutional Neural Networks (CNN), in 1D and 2D, are used on heart sound data, as a test case. The initial results show that the classification accuracy for the new data, with a higher SNR, coming into the 1D CNN is almost as high as if the network had been trained on the higher SNR data. For a 2D CNN working with spectrograms instead of time series data, the change in accuracy is not nearly as high, as the 2D CNN model seems more robust to noise differences. 

Most biopolymers used as additives for the improvement of expansive subgrade soils are ecofriendly but highly uneconomical and unsustainable. Even the traditional additives such as cement, lime, and fly ash that are used widely for most soil improvement schemes are highly notorious for their carbon footprint. This necessitated the motivation in the present study to utilize an economical, ecofriendly and highly sustainable biopolymer, known as pregelatinized corn starch (PGCS), to improve the strength properties of an expansive subgrade soil. The PGCS was admixed with quarry dust (QD), an industrial waste additive, before blending with the expansive subgrade soil in different mix ratios generated with a 32 full factorial design experiment. The California bearing ratio (CBR) samples were subjected to 7 day curing while that of the unconfined compressive strength (UCS) were subjected to 1, 7, and 28 day curing. Shortly after the improvement of the expansive subgrade soil, the PGCS and QD were used as predictors in the development of two regression models for the two strength parameters (CBR and UCS) of the expansive subgrade soil considered in the study. Next, multiobjective salp swarm optimization algorithm (MOSSA), a bioinspired algorithm, was employed to optimize the additives in order to obtain optimal values of the strength properties of the expansive subgrade soil blended with the additives. The developed models were set as fitness functions in the slightly modified MOSSA technique. Thereafter, nondominated solutions were determined after the implementation of the optimization analysis. The results obtained from laboratory experiments and the optimization process showed that there was significant improvement in the UCS and CBR of the expansive subgrade soil. Optimal improvement in the UCS (1,326.241 kN/m2) and CBR (36.8%) were observed when an optimum mix ratio of the additives, 0.3117% PGCS and 10% QD, was blended with the expansive subgrade soil.

The present study aimed to assess the potential of a bio-inspired algorithm, multi-objective grey wolf optimization algorithm (MOGWO), to optimize the strength properties (California bearing ratio (CBR) and unconfined compressive strength (UCS)) of an expansive subgrade soil. This optimization process involves the use of two additives, namely a bio-polymer, pregelatinized corn starch (PGCS), and a nanoparticle agro waste, rice husk ash (RHA), blended with the soil in different mix ratios determined by a 32 factorial experimental design. The CBR samples were cured for 7 days, while the UCS samples were cured for 1, 7, and 28 days. To optimize the expansive subgrade soil strength, regression models were developed using PGCS and RHA as predictors for CBR and UCS, serving as fitness functions in the slightly modified MOGWO optimization technique. Next, the optimization analysis produced non-dominated solutions. The results obtained from the laboratory experiments and optimization analysis revealed that there was significant improvement in the UCS and CBR of the soil. These improvements can be attributed to the pozzolanic reaction between the soil-RHA matrix, the formation of intercalated and exfoliated nanocomposites, and the hydrophilic interaction of PGCS. By applying the slightly modified MOGWO technique, the study achieved optimal enhancements in UCS (710.3 kN/m2) and CBR (24.2%) when the expansive subgrade soil was mixed with 0.2637% PGCS and 12.2413% RHA. The results demonstrate the potential of the MOGWO technique in improving the properties of expansive subgrade soil. 

In cardiac auscultation, the ability to clearly hear any existing murmur sounds in heart sounds is crucial for proper diagnosis. This work aims to improve heart sound by the use of a stethoscope array and beamforming technique. The stethoscope array comprises four piezo elements for measurement, placed on the edges of a 40mm by 40mm rectangle. The directionality of the piezo elements reduces the effect of ambient noise in the measurement. The signal amelioration is achieved by isolating the systole and diastole sounds, and independently applying the delay-and-sum beamforming. This thereby makes any existing murmur sounds in the systole and/or diastole more audible and clearer to aid diagnosis. Finally, the designed stethoscope array and signal processing shows a gain of up to 33% for measured healthy heart samples, and up to 63% increase in murmur sound gain for measured sample with medically confirmed murmur. 

The joint and marginal probability densities of multipaths’ angles-of-arrival (AOA) and times-of-arrival (TOA) at the cellular base station are developed in closed form in this paper. Unlike the general simplification assumption in the open literature in which the scatterers are assumed to be located in a circular region for non-uniform spatial densities, the scatterers in this paper are assumed to be located in an elliptical region to properly model the elliptical footprint around the mobile station from the elevated base station with directional antenna. The inverted elliptic–parabolic spatial density was adopted to model the non-uniform distribution of the scatterers around the mobile. The uplink’s AOA–TOA joint distributions, AOA and TOA marginal distributions were analytically derived in closed form. How the eccentricity of the elliptical scatterer region affects the probability density functions is discussed. Furthermore, the derived AOA marginal distribution is compared to that of the elliptic conic and inverted parabolic models. The proposed model is shown to have better least-squares fit to some empirical AOA data compared to the elliptic conic and inverted parabolic models.

Traditional spherical sector microphone arrays using omnidirectional microphones face limitations in modal strength and spatial resolution, especially within spherical sector configurations. This study aims to enhance array performance by developing a spherical sector array employing first-order cardioid microphones. A model based on spherical sector harmonic (SSH) functions is introduced to extend the benefits of spherical harmonics to sector arrays. Modal strength analysis demonstrates that cardioid microphones in open spherical sectors enhance nonzero-order strengths and eliminate the nulls associated with spherical Bessel functions. We find that the spatial resolution of spherical cap arrays depends on the array’s maximum order and the limiting polar angle, but is independent of the microphone gain pattern. We assess direction-of-arrival (DOA) estimation performance for coherent wideband sources using the array manifold interpolation method, and compare cardioid and omnidirectional arrays through simulations in both open and rigid hemispherical configurations. The results indicate that cardioid arrays outperform omnidirectional ones on DOA estimation tasks, with performance improving alongside increased microphone directivity in the open hemispherical configuration. Specifically, hypercardioid microphones yielded the best results in the open configuration, while subcardioid microphones (without nulls) were optimal in rigid configurations. These findings demonstrate that spherical sector arrays of first-order cardioid microphones offer improved modal strength and DOA estimation capabilities over traditional omnidirectional arrays, providing significantly enhancing performance in spherical sector array processing.