Underwater acoustic communications (UAC) in shallow waters is an extremely challenging task. This becomes even more difficult when reliable communications with an object buried in seabed sediments is required. The work presented in two articles (Part A and Part B) is a continuation of the earlier work published in [1]. The articles present the simulation of an acoustic transmission channel, taking into account the bottom slope angle. The simulation utilize the impulse response method and the image source ray tracing technique. The articles present examples of data transmission in a shallow channel with a sloping bottom to a receiver submerged in the water column (Part A) and to a receiver buried in seabed sediments (Part B).
Repetitively pulsed lasers operating in the MWIR, LWIR and FIR spectral regions are increasingly used in a wide range of applications, particularly in the form of semiconductor sources valued for their stability, efficiency and compactness. Ensuring the safe use of such devices requires accurate classification in accordance with IEC/EN 60825-1 standard, a process that becomes complex when multiple emission parameters such as pulse duration, repetition rate and peak power vary simultaneously. Although the standard provides general guidance and limited examples, it does not fully address all possible parameter combinations encountered in practical systems. This work presents a simplified analysis of the safety classification of repetitively pulsed lasers with variable parameters. A MATLAB-based simulation tool was developed to explore classification outcomes within defined parameter limits and to generate classification maps as a function of peak power and pulse repetition rate. The approach offers a clearer understanding of how change of parameters influences classification and provides a foundation that can be extended to broader parameter ranges and more diverse laser systems.
Anomaly-based network forensics is very importantfor finding new types of cybercrime that don’t have reliablesignatures or labelled training data. But most unsuperviseddetectors only look at one view of normality and don’t giveany forensic interpretability. This study talks about E-HUNF,an Explainable Hybrid Unsupervised Framework that can findcrimes in network traffic. E-HUNF uses a manifold-aware, centreregularizedauto encoder to get compact latent representationsof flows. It then uses these to get three different anomaly scoresbased on reconstruction error, latent density, and distance from alearnt normalcy centre. These scores are combined into a hybridanomaly score with adaptive, percentile-based thresholding tohelp people make judgements that are mindful of risk. Anexplainability layer blends local linear surrogates with prototyperetrieval to show how each alert’s features and historical examplesare related. When tested on a standard network-forensicsdataset with benign, DoS, Probe/Scan, R2L/U2R, and Botnettraffic, E-HUNF got an accuracy of 0.987, an F1-Score of 0.978,a ROC-AUC of 0.995, and a PR-AUC of 0.993. It did betterthan Deep SVDD, DAGMM, VAE-AD, and Isolation Forest. Evenfor small R2L/U2R attacks, the class-wise F1-Scores stay above0.937. Ablation results show that adding density and boundarycues to reconstruction improves the F1 score by 3.3% overreconstruction-only versions. These results show that E-HUNFhas the best detection performance and the most useful forensictransparency for modern cyber-defence operations.
This work presents the fabrication process and I–Vcharacteristics of ISFET devices with an open gate. Measurementswere performed in deionized water and in deionized watersolutions containing iron(II) fumarate at various concentrations.Changes in the electrical parameters of the transistors wereanalyzed as a function of electrolyte composition, with particularemphasis on the influence of Fe2+ ions and fumarate anions. Thestudy aimed to assess the sensitivity and stability of ISFET devicesin a complex ionic environment.
This research paper aims to develop aerial mappingbased on visual using RTAB-MAP for search and rescue robotapplications. Mapping these areas is crucial for locating disastersites that are challenging for humans to reach. Area mappingutilizes a visual RGB-D camera as input and RTAB-Map as a dataprocessing system to represent a 3D map. RVIZ ROS is used forvisualizing the map in 3D format. RGB-D images are processedusing RTAB-Map to obtain several data points, includingodometry, loop closure, timestamp, x, y, z coordinates, and roll,pitch, yaw data from the camera. The test results yielded anoptimal area mapping, where the loop closure results in findingcorresponding image data quite effectively. As the number of loopclosures increases, the area mapping quality improves. In thisstudy, there were a total of 450 loop closure data, which allowedthe system to create an optimal 3D map representation. Thecreation of this map is expected to assist in visualizing hazardousareas to aid in search and rescue efforts during natural disasters.
Nowadays, the trend in communication technology isgathering clothing with electronic technology. There are twofactors despite each others, the convenient of On-Body antennaspower supply on the cloth and the power combined with thebiological lossy tissues for the human; therefore, efficiency isimportant when considering Body-Centric communicationsystems. In this work, the experiments conducted in the study ofOff-Body radio with water in the channel are described. Tomanage the searching properties of water two techniques werestudies. The first was a simulation of multilayeres human modelwith and without a water layer in aspect of transmissioncoefficient and antenna reflection, the second was to use a gel toabsorb water, Polyacrylamide, within special box underneath theOn -Body antenna measuring the s11 in frequency 2.1 GHz
This study proposes a new intelligent framework tocope with the challenges involved with dynamic resource allocationin the 5G network environment based on Proximal PolicyOptimization (PPO), which is one of the most successful DeepReinforcement Learning (DRL) techniques. We have reformulatedresource allocation as a Markov Decision Process (MDP). Here, the"state" represents the current status of the network in terms ofdemand, interference, and channel quality. At the same time, the"Action" represents the allocation decision made for each serviceslice in terms of spectrum, capacity, and time. The proposed modelfocuses on balanced dynamic resource allocation across three mainsegments: eMBB, URLLC, and mMTC, through ensuring thatQoS requirements for each segment are met without impact to theoverall system performance. Our simulation results havedemonstrated excellent performance by the proposed algorithmwhen compared to traditional algorithms (i.e., GA, PSO, QLearning,and Round Robin). In our results, we showed athroughput increase of approximately 180 Mbps, energy efficiencyof 0.91 bps/joule, a Fairness Index of 0.88 overall performanceimprovement between 12% to 15%. As a result of the simulationresults, we believe that the PPO-MDP Framework is a good,realistic option for optimizing the use of resources within adynamically segmented environment, thus improving the ability ofa 5G system to efficiently and sustainably respond to a variety ofservice demands.
Radio Frequency Identification (RFID) is a coretechnology inside the rapidly expanding Internet of Things (IoT),with several applications in various industries. However, RFIDsystems have significant privacy and security risks. To addressthese difficulties, this paper proposes an efficient RFID-basedmutual authentication protocol (MAP) that uses the PRINCElightweight cipher. The proposed PRINCE cipher utilizesa pipelined design supported by an enhanced control mechanismfor round operations, resulting in significantly improvedthroughput and operating efficiency. This design's integratedhardware architecture allows it to carry out both encryption anddecryption duties smoothly. The protocol uses five encryption andtwo decryption operations during the mutual authenticationprocess between RFID tags and readers to provide a safeconnection. Performance evaluations show that the suggestedPRINCE cipher outperforms existing solutions in terms of areautilization and efficiency. Operating at 224 MHz, the architecturehas a remarkable latency of 3.5 clock cycles and a throughput of4.11 Gbps. When integrated into the RFID-based MAP, it achievesa throughput of 374.6 Mbps, an efficiency of 140.7 Kbps/Slice, anda processing latency of 0.355 μs. Such performance shows itseffectiveness and suitability as a robust security solution for IoTenvironments.
A wireless sensor network is vital in various fieldsand entails of a large number of sensor nodes. These nodesperform multiple functions, such as sensing, processing,communication, and power management. In such networks, alarge amount of data is generated and transmitted from sensornodes to the destination. During data transmission, congestionmay occur at the cluster head as well as during communicationbetween sensor nodes. This congestion mainly arises due toinefficient resource allocation and uneven traffic distribution. Themain reason for packet loss and the need for retransmission aredue to traffic imbalance and congestion. To avoid suchcircumstances an efficient flow control and congestion controllingschemes need to be adapted. To manage congestion control, alimited number of wireless sensor networks with differentprotocols are employed. The Deterministic Energy EfficientClustering (DEC) protocol, which relies entirely on residualenergy, is considered to minimize energy consumption. Thisprotocol, along with the leaky bucket algorithm, is used toregulate data flow at the cluster head and effectively controlcongestion in the network. This idea offers a solution to addresstraffic congestion and makes improvements to the situation in theevent that it arises. Simulation results show that the definedmethod can considerably advance increases in lifetime, energyand throughput.
This paper investigates the robustness of traffic signclassification models against real-world visual disturbances. Weconduct a comparative evaluation of three distinct architectures:a standard CNN, a hybrid CNN enhanced with Kolmogorov-Arnold dense layers (CNN-KAN), and a fully convolutionalKolmogorov-Arnold Network (CKAN). Unlike traditional CNNs,the KA-based models utilize learnable activation functions, potentiallyoffering improved resilience. The experiments were conductedusing the German Traffic Sign Recognition Benchmark(GTSRB) dataset, containing 43 classes of traffic signs. Modelswere trained and tested on both original images and versionsdegraded by controlled disturbances, including rotation, blur,brightness variation, and simulated rain. The results demonstratethat the proposed CNN-KAN model provides consistently superiorperformance under small-to-moderate rotations (up to 20degrees) and moderate brightness increases, achieving the highestaccuracy in all rain-mask scenarios. It remains competitiveunder blur, where it ranks second only to the standard CNN.Performance decreases were observed only at extreme brightnesslevels, where both the standard CNN and CKAN maintainedhigher stability. Overall, the findings highlight the potential ofKolmogorov-Arnold-based architectures for improving robustnessin traffic sign recognition systems operating under realisticand dynamically changing environmental conditions.
Passive radar does not transmit its own signals butrelies on external sources of illumination. Estimating a targetposition requires distance measurements from at least threespatially separated sensors. This paper presents an adaptation ofa maximum likelihood estimator (MLE) for target localizationusing multiple (three or more) bistatic range or DToA (DifferenceTime of Arrival) measurements. The Gauss–Newton, Levenberg,and Levenberg–Marquardt methods are compared with respectto numerical stability and localization accuracy under differentnoise levels and initialization conditions.
The foreign exchange (Forex) market is highly liquidand volatile, making accurate short-term forecasting both criticaland challenging. This study investigates one-day-ahead AUD/USDexchange rate prediction using CatBoost, Random Forest (RF),and Support Vector Machine (SVM) machine learning (ML)models with continuous and discretized technical indicators. Tentechnical indicators were derived from 5,027 historical data points.This is the first study to apply discrete technical indicators withCatBoost for recent AUD/USD price forecasting. Results showedthat CatBoost achieved the highest accuracy (89.68%) and AUC(0.9609) on the discretised dataset. Statistical test confirmed thesignificance of CatBoost’s superior performance, highlighting itspotential to enhance predictive performance and support real-timedecision-making in Forex trading.
This article investigates the feasibility of upgradingWi-Fi modules in older PC laptops to comply with modern IEEE802.11 standards, aiming to enhance network performance.Experimental results show that modern Wi-Fi adapterssignificantly improve throughput and efficiency, particularly in the5 GHz band. However, challenges such as driver availability, BIOSwhitelists, and hardware compatibility must be taken into account.The findings suggest that Wi-Fi upgrades can extend the usabilityof older laptops cost-effectively, with future research planned toexplore newer standards and configurations.
This paper investigates the integration ofreconfigurable intelligent surfaces (RIS) mounted on cooperativeunmanned aerial vehicle (UAV) swarms to enhance mmWavecommunication in dense urban environments. While mmWavebands offer large bandwidths for high data rates, their severe pathloss and blockage remain major obstacles in practicaldeployments. UAVs provide flexible line-of-sight (LoS)connectivity, whereas RIS enables programmable reflection tostrengthen non-line-of-sight (NLoS) links. By combining thesetechnologies, we develop a proposed framework that determinesUAV positions and RIS phase configurations under realisticconstraints, including quantized RIS phases, UAV, and flightrestrictions. Simulation results demonstrate that RIS-aided UAVswarms significantly improve system reliability compared tosingle-UAV or non-RIS baselines. The findings reveal that bothUAV swarm size and number of RIS elements strongly influenceperformance: average signal-to-noise ratio (SNR) increases withmore UAVs, bit error rate (BER) decreases with RIS enlargement,and outage probability exhibits a steep once the number of UAVsand RIS elements is reached a certain limits. In particular, swarmsof four to five UAVs with 16–20 RIS elements achieve outageprobabilities below 0.05, ensuring near–ultra-reliablecommunication while balancing energy and hardware costs. Theseresults confirm that cooperative RIS-enabled UAV swarms are apromising solution for robust mmWave coverage in urban 6Gnetworks.
The advanced GaN-based devices utilized for highperformanceradio frequency (RF) applications are intensivelystudied to be used as RF sensors or amplifiers. The paper is focusedon microwave characterization of two port passive devices,especially on-chip calibration structures and InAlGaN/GaNelectron mobility transistor (HEMT) operating in the cold biasregion (zero applied voltage). The acquired S-parameters areinputs to build a passive device small-signal model consisting ofthree star-connected impedances. The calculated Z-parameters arepossible to be utilized for on-chip signal paths design. Theparameters to be calculated are assumed frequency independent,however, more proper HEMT modelling requires non-zero voltageapplication, therefore, the model possibilities are depicted anddiscussed.
This paper presents a text-independent speakeridentification system that utilizes MFCC, LPC, prosody, andoptimized multi-level DWT features for robust speaker modeling.The system is designed for multiple standard speech databases,including TIMIT, NTIMIT, SITW, and NIST2008. During training,features from each speaker are normalized to zero-meanand unit-variance, and Student-t distributions are fitted to modelthe statistical characteristics of each speaker. For testing, featuresare normalized using the corresponding speaker’s trainingstatistics, and speaker identity is predicted based on maximumlog-likelihood estimation over the trained models. Experimentalresults confirm the superiority of the proposed system, whichachieves high accuracy across multiple datasets (e.g., 98.33% onTIMIT, 89.38% on NTIMIT, 96.88% on SITW, and 100.00%on NIST2008) and consistently outperforms existing state-of-theartmethods under AWGN conditions, demonstrating significantimprovements in identification accuracy and the effectiveness ofmulti-feature fusion and Student-t modeling.
This paper presents a review of recent scientificliterature concerning the use of cocotb as a Python-based frameworkfor functional verification of digital systems. The studycategorizes existing works into three groups: design verificationcase studies, tool enhancement methodologies, and comparativeanalyses between cocotb and SystemVerilog/UVM environments.The strengths and limitations of cocotb are evaluated with respectto accessibility, ecosystem maturity, constrained random verificationcapabilities, and industrial applicability. The analysis revealsthat cocotb provides a flexible, cost-effective solution, particularlysuited to early-stage RTL verification and open-source workflows,while SystemVerilog/UVM remains advantageous for large-scaleindustrial projects due to its mature ecosystem and commercialtool integration. The paper identifies current gaps in methodologyevaluation, coverage analysis consistency, and reproducibility inexisting research, and outlines directions for future developmentof hybrid verification flows.
Underwater wireless sensor networks are widely usedin sea and ocean exploration, monitoring of the environment,defense surveillance. These applications are restricted by limitedenergy availability, propagation delay of acoustic signal, andtopology changes. To address these issues, a reinforcementlearning (RL)-based routing protocol that combines energy-awareclustering with Q-learning to improve packet forwardingefficiency is proposed in this paper. In this approach, the role ofeach autonomous agent is performed by sensor node andforwarding actions based on residual energy, hop count, anddistance to the sink are adaptively selected. MATLAB simulationresults demonstrate that the proposed scheme achieves a packetdelivery ratio (PDR) of 95.2%. Compared with vector-basedforwarding (VBF) and reinforcement learning-based opportunisticrouting (RLOR), the achieved PDR is 7.6% and 3.7% higher,respectively. The improvement of performance is mainlyattributed to adaptive Q-learning-based next-hop selection andenergy-aware clustering, which reduce redundant and longdistancetransmissions and avoid routing voids. Moreover, theproposed protocol extends network lifetime to 5000 iterations,achieving improvements of 19% and 6.4%, while reducing averageenergy consumption by 25.7% and 13.3% compared with VBF andRLOR.
Phishing is widely acknowledged as one of the mostinsidious types of social engineering attacks. Despite substantialefforts to combat this issue, it continues to evolve in sophistication,resulting in increasing financial losses. Historically,countering phishing involved a blend of human vigilance andsoftware-based detection mechanisms, primarily relying on listbasedstrategies. However, with the advent of advanced datascience, innovative phishing detection techniques utilizing MachineLearning models have emerged and garnered significantresearch attention. This study aims to comprehensively comparethe effectiveness of traditional heuristic-based and modern MachineLearning classification models, while addressing challengesassociated with their efficiency. Experimental results involvingthe Random Forest classifier, although requiring slightly morecomputational power, demonstrated a substantial increase indetection accuracy (57.2% higher) and a remarkable reduction intesting time (11.28 seconds faster vs 0.01 seconds) when comparedto heuristics using the same input data.
Autonomous robots are expected to be crucial inmitigating the lack of precise border surveillance in complexterrains, especially within forests, mountains, and industrialsettings, focused on predicting the entry of terrorists and illegalrefugees. These robots must demonstrate navigational proficiencyin both controlled areas, such as border control gates andborders with plane surfaces, and unplanned country borders indaily surveillance tasks, including mountains, forests, and mud.In this regard, quadruped robots are attracting considerableinterest due to their ability to carry substantial surveillanceequipment while navigating inclines and obstacles. Given theenvironment-dependent characteristics of optimal gait patterns,it becomes a necessity for mechanisms capable of independentlyand effectively optimizing gait patterns for adaptation to differentscenarios. To address these challenges, we explored a model forquadruped locomotion that combines central pattern generators(CPGs) with deep reinforcement learning (DRL). Using bothexternal and internal sensing, the agent learns to coordinaterhythmic movement among multiple oscillators to follow speedinstructions, while also adjusting these instructions to avoidbumping into things around it. This research helps to useDRL to study important questions in neurology, such as howcertain pathways, connections between oscillators, and sensoryinformation affect walking patterns. Moreover, this paper willpresent fundamental knowledge, research trends, and challengesregarding the implementation of the CPG-DRL framework inrobotic perception.
