Specific Emitter Identification (SEI) authenticates a transmitter from the hardware-intrinsic fingerprint it leaves on its signal, which makes it attractive for physical-layer security. The difficulty is that these analog impairments are faint, and in high-mobility links they must be read through severe time-selective Rayleigh fading with Doppler. Real-valued deep networks make this harder than it needs to be: splitting the in-phase and quadrature branches weakens the joint phase-amplitude coupling, and cross-entropy training leaves decision boundaries that drift as the channel changes. This paper takes a different route with a Complex-Valued Transformer (CV-Trans) trained under a Supervised Complex Contrastive Learning (SupCon) objective. The model stays in the complex domain throughout: it fuses the raw envelope with differential-trace phase features to retain the In-phase/Quadrature (I/Q) mismatch and phase-noise signatures, and the SupCon loss tightens each emitter's cluster on a unit hypersphere so that the fingerprint separates from stochastic Doppler rotation. Under Rayleigh fading with Doppler, the framework reaches 99.97% accuracy at 20 dB SNR and holds 99.7% at high mobility ( fdTs=10-2 ), while using 50% fewer trainable linear parameters than the real-valued equivalent and running in about 2.73 ms per burst - fast enough for real-time use.
In this study, a systematic method is presented for planning the network reconfiguration strategy to identify the most key switches or critical switches. This is to achieve a good voltage profile for distribution operation purposes. The optimal switching criterion is developed using permutation importance, a machine learning technique. Standard deviation is used to analyse the distribution of the load flow dataset. Permutation importance is a machine learning technique used to measure the contribution of an input to the output. Whereas standard deviation is a useful technique for measuring scattered data, it is used to identify values that are closer to the mean and aids in identifying the error margin of the outcomes. An overall load flow solution is carried out using the sample data of the average voltage profile of all the buses at different configurations under load-varying conditions. There may be a large number of feasible configurations based on switching sequences of the system, and out of that, some sample configuration data have been considered for testing purposes. This study presents a comparison between machine learning techniques and statistical models aimed at improving the performance of power networks to safeguard against irregular events.
This paper presents an explainable framework for peripheral oxygen saturation (SpO2) and the classification of hypoxemia levels using photoplethysmography (PPG) signals obtained from a MAX30102-based optical sensing dataset. The study utilizes the publicly available Hemoglobin PPG dataset, which includes only red and infrared channels and was initially created for hemoglobin estimation. Since there were no SpO2 annotations in that dataset, the red-infrared intensity ratios and standard oxygen saturation thresholds were modeled physiologically to provide synthetic labels. To create interpretable input features, the preprocessed signals were subjected to windowed segmentation, wavelet-based denoising, and AC/DC feature extraction. Several machine learning (ML) models were used for classification, including feedforward neural networks (FFNN), hybrid ensembles, Extreme Gradient Boosting (XGBoost), Extra Trees, and a stacking ensemble. Among these, Extra Trees achieved the best accuracy, at 99.09%. The need for interpretable, computationally effective deployment on edge devices and the small dataset size made traditional machine learning methods more appealing than deep learning. The physiological validity of important spectrum features was validated by Explainable Artificial Intelligence (XAI) modules using partial dependency analysis and permutation importance.
Nowadays, fossil fuels, such as petrol, diesel, and coal, are being replaced by non-conventional energy resources (NCERs) because NCERs are free from CO2 emissions, and their availability is surplus in the environment. Because of these major advantages, NCERs are the main focus for researchers. The hydro energy-based power system (HEPS) is one of the important NCERs. The modelling of HEPS is complicated by their higher-order nature. Consequently, the literature is progressively proposing reduction approaches. This study proposes a novel model order reduction approach to approximate the higher-order HEPS, based on the particle swarm optimisation (PSO) algorithm and the factor division technique (FDT). The PSO algorithm is used to estimate the denominator polynomial, while the FDT approach is used to generate the numerator polynomial of the desired reduced order model (ROM). A sixth-order HEPS model is considered as a test case to demonstrate the superiority of the suggested approach. The outcomes are contrasted with established techniques found in the literature.
