Indoor Positioning Systems (IPS) are essential for delivering accurate location-based services in environments where Global Navigation Satellite Systems (GNSS) are ineffective. The Received Signal Strength Indicator (RSSI)-based IPS leverages the existing access point infrastructure to provide a cost-effective solution for indoor location determination. However, device heterogeneity, characterized by variations in hardware, sensors, and software between devices, poses a significant challenge, often degrading positioning accuracy and robustness. This study investigates the impact of device heterogeneity on IPS performance using the TUJI1 dataset, which comprises RSSI measurements collected from five different devices. After comprehensive preprocessing of the RSSI signals, the study proposes a hybrid Convolutional Neural Network–Long Short-Term Memory (CNN-LSTM) framework, leveraging advanced feature extraction techniques to improve positioning accuracy and mitigate the effects of device variability. The proposed model achieves a three-dimensional (3D) mean positioning error of 2.20 meters, outperforming traditional k-Nearest Neighbors (k-NN) methods. In cross-device evaluations, the model demonstrates improved robustness, reducing positioning errors by up to 0.17 meters compared to conventional approaches. These results highlight the effectiveness of the CNN-LSTM architecture in addressing device heterogeneity, offering a scalable and efficient solution for diverse IPS environments. This study advances the field of indoor positioning by providing a robust framework capable of maintaining high accuracy in the presence of device variability, thus contributing to the development of more reliable and adaptable IPS technologies.

Motor imagery (MI) is a widely used paradigm in brain–computer interface (BCI) systems, with applications in rehabilitation and neuroscience. In this study, magnetoencephalography (MEG) signals were employed to analyze MI and other mental imagery tasks. MEG provides high spatial resolution, facilitating the classification of imagery-related signals. This study aims to enhance the classification of motor and cognitive imagery (CI) tasks using a public MEG dataset including four distinct tasks: imagining the movement of hands (H) or feet (F), performing arithmetic subtraction (S), and forming words (W). MEG signals were decomposed using five signal-decomposition methods: Empirical Wavelet Transform (EWT), Maximal Overlap Discrete Wavelet Transform (MODWT), Empirical Mode Decomposition (EMD), Variational Mode Decomposition (VMD), and Multivariate Variational Mode Decomposition (MVMD). Feature extraction was performed using the Common Spatial Patterns (CSP), with t-test-based feature selection. Subsequently, commonly used machine learning algorithms were employed to classify MI and CI tasks. The results indicate that MVMD and MODWT achieved the highest accuracies when combined with the Artificial Neural Networks. MVMD yielded superior performances in (H and W: 79.2%; F and S: 75.8%; and F and W: 73.8%) tasks. MODWT achieved high accuracies in the H and W (75.9%) and F and W (76.3%) tasks. Overall, motor and non-motor pairs (H and W, F and W) yielded higher accuracy than the cognitive pair (W and S).

Autism spectrum disorder (ASD) is a neurodevelopmental condition characterized by impairments in social interaction and communication, along with atypical behavioral patterns. Affected individuals often seem isolated in their inner world and exhibit particular sensory reactions. The World Health Organization has indicated a persistent increase in the global prevalence of autism, with approximately 1 in 127 persons affected worldwide. This study contributes to the growing research effort by presenting a comprehensive analysis of functional connectivity patterns for ASD prediction using rs-fMRI datasets. A novel approach was used for ASD identification using the ABIDE II dataset, based on functional networks derived from BOLD signals. The sparse functional brain connectome (Lean-NET) model is employed to construct subject-specific connectomes, from which local graph metrics are extracted to quantify regional network properties. Statistically significant features are selected using Welch’s t-test, then subjected to False Discovery Rate (FDR) correction and classified using a Support Vector Machine (SVM). Our experimental results demonstrate that locally derived graph metrics effectively discriminate ASD from typically developing (TD) subjects and achieve accuracy ranging from 70% up to 91%, highlighting the potential of graph learning approaches for functional connectivity analysis and ASD characterization.