National Institute of Telecommunications

Aim: Meteorological factors pose a significant threat to food security through disruptions in food availability, access, utilization, and stability.Extreme temperatures, erratic precipitation patterns, and extreme weather events adversely affect agricultural productivity, fisheries, and livestock, exacerbating inequalities in food access.This study examines the impact of meteorological factors on food security in Pakistan using the Food Insecurity Experience Scale (FIES) and investigates how household-level adoption of Information and Communication Technology (ICT) mitigates these effects across different climatic zones. Material and methods: Using district-level climatic data from NASA Power and household-level ICT adoption indicators sourced from the Pakistan Social and Living Standards Measurement 2019-2020 survey, the study examines the impact of climate factors on food security and examines the moderating effect of ICT. The estimation is done at the national and climatic zones utilizing linear (OLS) and quantile regression. Moreover, the direct and indirect effects of meteorological factors are calculated, and the moderating role of ICT is computed.Results: The findings reveal that all four meteorological variables negatively and significantly affect household food security.Among these, wind speed and precipitation exert the strongest adverse effects, particularly in tropical zones.The results also confirm regional heterogeneity, with the tropical zone being the most vulnerable.The quantile regression shows that the mitigation through ICT is most effective for households with a low level of food security.

This work presents the design and experimental validation of a 2 × 2 MIMO communication system assisted by a directly modulated analog radio-over-fiber (A-RoF) fronthaul, targeting low-complexity connectivity solutions for underserved/remote regions. The study details the complete end-to-end architecture, including a wireless access segment to complement the 20-km optical fronthaul link. The system is implemented on an software defined radio (SDR) platform using GNU Radio 3.7.11, running on Ubuntu 18.04 with kernel 4.15.0-213-generic. It also employs adaptive modulation driven by real-time signal-to-noise ratio (SNR) estimation to keep bit error rate (BER) close to zero while maximizing throughput. Performance is characterized over 20 km of single-mode fiber (SMF) using coarse wavelength division multiplexing (WDM) and assessed through root mean square error vector magnitude (EVMRMS), throughput, and spectral integrity. The results identify an optimum radio-frequency drive region around −16 dBm enabling high-order modulation (e.g., 256-QAM), whereas RF input powers above approximately −10 dBm increase EVMRMS due to nonlinearity in the RF front-end/low-noise amplifier (LNA) and direct modulation stage, forcing the adaptive scheme to reduce modulation order and throughput. Over the optical-power sweep, when the incident optical power exceeds approximately −8 dBm, the system reaches ∼130 Mbps (24-MHz channel) with EVMRMS approaching ∼1%, highlighting the need for careful joint tuning of RF drive, optical launch power, and wavelength allocation across transceivers. Finally, the integrated access link employs diplexers for transmitter/receiver separation in a 2 × 2 configuration with 2.8 m antenna separation and low channel correlation, demonstrating a 10 m proof-of-concept range and enabling end-to-end spectrum/EVM/throughput observations across the full communication chain.

Introduction: Neurological injuries significantly impair quality of life by disrupting neural transmission. Traditional implantable stimulators often rely on internal batteries, which limit device longevity and necessitate repeated surgical interventions. Objective: This study presents the experimental validation of a battery-free, RFID-powered neural platform for peripheral nerve signal acquisition and stimulation, targeting TRL-6 validation. Methods: The prototype incorporates an adjustable analog front-end with gains up to 93 dB and a biphasic current-controlled stimulator. Validation was performed through benchtop testing, biological tissue assessments using porcine tissue, and functional in vivo trials in adult Wistar rats (n = 3) over a three-month period. Results: Benchtop evaluation confirmed gain accuracy with errors below 2.2 dB and precise stimulation timing. The system maintained a stable 3.3 V wireless power link through 20 mm of biological tissue using RFID. In vivo experiments indicated a 100% functional success rate (51/51 trials) in eliciting gross motor responses via wireless stimulation. Thermal safety was confirmed, with a maximum operating temperature of 28 °C, remaining well below physiological limits. Conclusions: The results demonstrate the functional feasibility of a battery-free, RFID-powered neural interface for wireless signal acquisition and stimulation, supporting system-level validation of this architecture.

The rapid growth of IoT devices demands scalable and efficient threat detection solutions. This paper introduces a sample-based, multi-stage machine learning (ML) pipeline for IoT threat detection using the CICIoT2023 dataset, integrating feature selection, data balancing, and hyperparameter optimization to improve detection accuracy while reducing the computational overhead associated with training. We evaluate, and across binary, multiclass, and fine-grained tasks, showing that with 10% sampling achieves the best trade-off between accuracy and efficiency. Compared to prior methods, our approach eliminates GPU dependence, maintains low latency, and preserves state-of-the-art performance while enabling scalable training for high generalization capacity. Additionally, we provide model selection guidelines based on dataset complexity and computational constraints. The results show that training with a sample-based approach enables effective threat detection on large datasets, producing models that generalize well to diverse IoT attack scenarios, thus ensuring practical applicability in real-world deployments.

The integration of flexibility, electrical conductivity, and cost-effectiveness remains a central challenge in the fabrication of metasurfaces for electromagnetic applications. Here, we introduce a hybrid pen-based and mask-assisted fabrication strategy for Ti3C2Tx MXene-based metasurfaces on paper substrates, enabling precise patterning with minimal material waste. The approach combines screen-printing-derived adhesive masks with a direct ink writing process using MXene inks, yielding conformable and conductive metasurfaces that operate efficiently across the microwave range. Atomic force microscopy and scanning electron microscopy analyses confirmed the formation of uniform, ultra-smooth MXene films (root-mean-square roughness ≈438 pm, with a thickness of ≈13μm) atop paper layers with a total device thickness of 163.6μm. Experimental transmission spectra revealed strong stopband resonances for metasurfaces based on square-ring and split-ring resonators, with reflection dips near −12 dB. These results are consistent with finite-element method simulations performed using COMSOL Multiphysics, confirming the consistency between numerical predictions and laboratory measurements. The main limitation we observed was the moderate conductivity of our prepared MXene films (8.97×103 S/m), which can be enhanced through ink optimization. The low-cost, scalable, and sustainable MXene patterning shown in this work opens new opportunities for fabricating lightweight, flexible, and reconfigurable electromagnetic devices, particularly in wearable and next-generation wireless communication systems.

Background and aims – During ongoing morphological studies of Adenocalymma , we realized that nomenclatural updates are necessary in the genus. Here, we propose a new circumscription of A. acutissimum and the description of the new species A. darwinii . Material and methods – Morphological and phylogenetic evidence support the new circumscription of A. acutissimum , while the new species is recognized based on morphological grounds and compared to A hirtum and A. salmoneum . Key results – We removed the taxa A. comosum , A. comosum var. lanceolatum, and A. nitidum as synonyms of A. acutissimum . We also presented a detailed description for A. darwinii sp. nov., an illustration, and information on its distribution, habitat, phenology, and extinction risk. A map showing the occurrence of both A. comosum and A. darwinii sp. nov. is shown. Conclusion – These results emphasize once again the importance of taxonomic studies in the Brazilian Atlantic Forest and highlight conservation concerns for its flora.

Driver drowsiness and distracting behaviors are leading causes of road accidents. This paper presents a cost-effective, real-time driver monitoring system (DMS) that leverages Edge Machine Learning (EdgeML) to detect multiple risk factors: drowsiness, phone usage, smoking, and seatbelt non-compliance. The proposed solution integrates an ensemble of models, object detection (OD), facial landmark analysis, and posture estimation, deployed on a Raspberry Pi 5 (RPi 5) with a Coral USB Edge TPU (ETPU) accelerator. A dual-validation logic cross-confirms biometric anomalies with OD within a temporal window to minimize false positives. The system achieves an average F1-score of 98.66% and 90.46% for two- and five-class classification, respectively, and supports real-time processing at 40 FPS with low resource utilization (21% CPU, 500 MB RAM). This work offers a retrofittable EdgeML solution for real-time driver monitoring, democratizing access to advanced safety features for the billions of cars already on the road. The source code and model weights are publicly available<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">1</sup>.

The classical passive coherent location (PCL) processing makes use of at least two antennas and is based on a calculation of cross-ambiguity function (CAF) between reference (REF) and surveillance (SURV) signals. In the modern approach, which is dedicated to communication signals with orthogonal frequency division multiplexing (OFDM) and known as an OFDM radar, the REF signal can be reconstructed from the SURV one. Then, the channel frequency response (CFR) is estimated, and information about moving targets is extracted from it. Such an approach enables building a simple, single physical channel passive radar, which requires relatively low computational power compared to the classical, two-channel CAF approach. This paper shows what signal processing operations should be added to a standard digital audio broadcasting (DAB) receiver to obtain the PCL functionality. Several field tests confirm the correctness of the described enhancement. Two low-cost software-defined radio (SDR) devices were used in the hardware setup, i.e., RSPduo and RTL-SDR, to detect airplanes of different sizes flying at various altitudes, in outdoor and indoor receive scenarios. The investigated DAB-based PCL radar's operating range has been also estimated based on a reference from ADS-B data. The opportunity of a single-channel PCL radar design, operating with minimum hardware and just one omnidirectional antenna, opens up the possibility of deploying a network of small and inexpensive PCL sensors.