Ricardo Llugsi

Automated Enigma cryptanalysis faces combinatorial challenges due to the avalanche effect, which destroys mathematical gradients and causes sequential deep learning architectures like Long Short-Term Memory (LSTM) to fail. This paper presents a hybrid methodology decoupling the search space of a three-rotor Enigma machine to avoid local minima. It combines parallel exhaustive search for rotor alignment with a heuristic AI algorithm based on Directed Gradient Ascent for plugboard deduction. Restricting mutations to erroneous characters using a crib reduces the combinatorial space exponentially. Executed via JupyterLab on an Intel i9 processor, the algorithm converged to historical 1940 keys in 3.6 s, proving high computational efficiency and cryptographic robustness.

Water contamination by petroleum products represents an increasing environmental concern due to its impact on aquatic ecosystems and water resources. This paper presents a low-cost electromagnetic reflectometry system for estimating the relative reflection coefficient of gasoline-contaminated water using a software-defined radio (SDR) and a pair of log-periodic antennas. The proposed system operates in the UHF band from 2.1 to 3.0 GHz using a continuous-wave measurement scheme to acquire the reflected signal from the liquid surface. To improve measurement stability and repeatability, the signal processing chain incorporates temporal alignment based on cross-correlation, residual phase error compensation, and coherent averaging of multiple signal acquisitions. Water samples with different gasoline concentrations were experimentally evaluated under a fixed measurement geometry, and the relative reflection coefficient was estimated from the amplitude and phase of the reflected signal. The experimental results demonstrate that the proposed methodology provides repeatable measurements and enables discrimination between different contamination ratios. The proposed approach represents a low-cost alternative for the electromagnetic characterization of liquids and provides a foundation for future environmental monitoring systems based on electromagnetic reflectometry.

Hollow-Core Fibers (HCFs) mitigate non-linear distortions but introduce severe asymmetric noise from surface scattering and signal-ASE beating, causing standard fixed-threshold receivers to fail. This study introduces a multi-band channel-aware adaptive detector utilizing Support Vector Machines (SVM) with Radial Basis Function (RBF) kernels. The classifier dynamically adjusts decision boundaries based on the active wavelength, achieving a 99.97% global bit recovery accuracy. In the C-band, no classification errors were observed within the finite testing subset, successfully falling below the Hard-Decision Forward Error Correction (HD-FEC) threshold without requiring physical receiver hardware modifications.

Hollow-Core Fibers (HCFs) enable low-latency communications but suffer from thermo-optic variations. Traditional multi-band numerical solvers are too slow for real-time management. This paper presents a machine learning-based digital twin using a Multi-Layer Perceptron (MLP) to predict physical degradation across O-band and C-band channels under thermal variations (−10 °C to 45 °C). Evaluated via a 70/15/15 data split on 10,000 spatial–thermal points, the model prevents overfitting and achieves a Mean Squared Error of 0.0145 and an R² of 0.9999. The digital twin executes in sub-milliseconds, enabling real-time intelligent management for next-generation telecommunication infrastructures.

This paper presents the implementation and field evaluation of a short-range frequency-modulated continuous-wave (FMCW) radar based on an ADALM-PLUTO software-defined radio, a 23 dBi dual-polarized antenna, and MATLAB processing. A 56 MHz triangular sweep centered at 5.5 GHz was transmitted continuously. Because transmission and frame acquisition were asynchronous, a two-period reference and cross-correlation were used only to locate the waveform boundary before full-period dechirping, Hann windowing, and a 131,072-point FFT. The receiver used manual gain of 57 dB. For single stationary vehicles from 10 to 30 m, the retained estimates yielded an MAE of 0.48 m, an RMSE of 0.52 m, and a mean absolute percentage error of 2.53%. Two-vehicle tests produced one dominant response at a common nominal range and distinct peaks at nominal separations of 10 and 15 m. These results support the practical suitability of the proposed SDR-based architecture for short-range distance estimation and multi-target detection under the evaluated conditions. Future studies will consider target separations closer to the theoretical resolution limit, repeated measurements, and controlled target positioning to provide a more detailed characterization of the system performance.

Réduction de l'incertitude dans les mesures environnementales à l'aide de l'estimation bayésienne et adaptative des moments : cas d'étude ville andine de Quito: Étant donné la topographie unique de Quito, il est difficile de prédire les changements climatiques dans cette ville. Cette thèse se concentre sur l'étude de données météorologiques, en particulier pour la ville de Quito. Pour atteindre cet objectif, des micro stations météorologiques automatiques (SMA) ont été installées sur des sites d'intérêt et les données ont été collectées sur un serveur FTP dans le Cloud en utilisant le réseau cellulaire disponible. L'objectif de cette étude est de prédire les paramètres environnementaux tout en identifiant les erreurs de mesure afin de fournir un ajustement de la calibration des stations pour corriger ces erreurs. Nous avons développé et analysé quatre modèles différents pour obtenir des prévisions fiables : un modèle ARIMA (autoregression intégrée à moyenne mobile), un modèle LSTM (mémoire à court et long terme), un modèle LSTM empilé et un modèle LSTM convolutif avec une réduction des erreurs d'incertitude. Pour détecter les erreurs dans les micro-stations météorologiques automatiques, nous utilisons des séries temporelles provenant de deux stations fortement corrélées avec la station en cours d'analyse pour obtenir une prévision du paramètre mesuré (température dans nos expériences) sur 24 heures. Cela nous permet de déterminer si la station en cours d'analyse enregistre des mesures erronées. Pour détecter les erreurs de mesure, une comparaison est effectuée de manière itérative avec les informations de chaque micro-station en fonction de ses stations voisines. La différence entre les coefficients de corrélation initiaux et ceux acquis au temps t est calculée. Si cette différence dépasse un certain seuil, l'algorithme signale une erreur et lance le calcul d'ajustement de cette erreur en se basant sur la prévision calculée pour cette station. Finalement, nous fournissons les paramètres de l'équation de calibration du capteur de la station en utilisant l'ajustement proposé. Pour mener cette étude, nous avons utilisé différentes techniques, notamment le calcul de coefficients de corrélation, l'utilisation d'un réseau de neurones multicouche, la conception d'une nouvelle version de l'optimiseur ADAM et une stratégie de réduction de l'incertitude basée sur une approche bayésienne.

This paper takes a fresh look at the influence of CO2 and CH4 for the identification of “El Niño” episodes. We first introduce the statistical methodology for early detection based on the analysis of the historical information Temperature Anomaly produced in the Oceans by the “El Niño” episode. Second, we introduce a statistical analysis to study a possible relation between CO2 and CH4 with the Temperature Anomaly. We apply three Neural Network models and we analyse the correlation between the obtained series and the real one as well as the respective error metrics. The outcome shows that a Convolutional Encoder-Decoder model is the most suitable structure to carry out this purpose because it shows a correlation coefficient of 0.991 and error metrics of MSE = 0.096, RMSE = 0.309, MAE = 0.249. However, the amount of historical information on CH4 becomes a limitation to detect a further relationship.

This paper presents a novel continuous-wave Doppler RADAR system for real-time speed measurement of moving objects, implemented using software-defined radio (SDR). Unlike traditional high-cost solutions typically found in research centers or specialized laboratories, this prototype offers a low-cost, compact, and easily deployable platform that lowers the entry barrier for experimentation and research. Operating within the 70 MHz–6 GHz range, SDR enables highly flexible signal processing; in this implementation, a 5.5 GHz carrier is selected to improve the detection precision by exploiting its reduced bandwidth for more accurate observation of frequency shifts. The carrier is modulated with a 2 kHz signal, and Doppler frequency deviations induced by object motion are processed to calculate velocity. Using a Welch spectral estimator, the system effectively reduces noise and extracts the Doppler frequency with high reliability. The prototype achieves speed measurements up to 196.36 km/h with approximately 2% error in the 0–100 km/h range, confirming its suitability for road traffic monitoring. A key innovation of this work is its single-antenna cross-polarized configuration, which simplifies hardware requirements while maintaining measurement accuracy. Furthermore, the system’s portability and open-access design make it ideal for in-vehicle applications, enabling direct deployment for automotive testing, driver-assistance research, and educational demonstrations. All design files and implementation details are openly shared, eliminating patent restrictions and encouraging adoption in low-resource academic and research environments.

This paper introduces a theoretical large-scale radio channel model for the downlink in cellular systems, aimed at estimating variations in received signal power at the user terminal as a function of device mobility. This enables applications such as direction-of-arrival (DoA) estimation, estimating power at subsequent points based on received power, and detection of coverage anomalies. The model is validated using real-world measurements from urban and suburban environments, achieving a maximum estimation error of 7.6%. In contrast to conventional models like Okumura–Hata, COST-231, Third Generation Partnership Project (3GPP) stochastic models, or ray-tracing techniques, which estimate average power under static conditions, the proposed model captures power fluctuations induced by terminal movement, a factor often neglected. Although advanced techniques such as wave-domain processing with intelligent metasurfaces can also estimate DoA, this model provides a simpler, geometry-driven approach based on empirical traces. While it does not incorporate infrastructure-specific characteristics or inter-cell interference, it remains a practical solution for scenarios with limited information or computational resources.

This study applied modified linear regression in machine learning (ML) to predict the direction of arrival (DoA) in cellular networks using field measurements and radiofrequency parameters. Models were developed from base station data, with preprocessing for pattern identification and formula adjustments to improve the accuracy across angle ranges. Machine learning, tested here as an additional method to traditional techniques, achieved a root mean square error (RMSE) of 3.63 to 17.93, demonstrating enhanced adaptability. While requiring substantial data and computational resources, this approach highlights machine learning’s potential as a valuable tool for DoA estimation in cellular networks.