This paper provides an extensive review of the role of 5G technologies in improving the performance of Mobile Ad-Hoc Networks(MANETs). The novelty of this study lies in the development of a Brown Boosted Expectation Maximization ensemble node clustering based energy-efficient and reliable data routing(BBEMENC) framework designed for 5th Generation(5G) enabled technologies on MANETs. By intelligently grouping nodes based on energy, trust, and signal metrics, and dynamically adjusting routing decisions, the model achieves substantial improvements in reliability and power efficiency. The study delves into key performance metrics, such as latency, throughput, scalability, and reliability, and explores how 5G enhances these aspects through advanced communication technologies like ultra-reliable low-latency communication(URLLC), network slicing, and edge computing. Furthermore, a comprehensive literature survey is conducted to analyze existing works on MANET performance under 4G and 5G environments. By employing simulations and real-world case studies, this review highlights the practical applications and potential challenges of integrating MANETs with 5G. The findings suggest that while 5G significantly improves network performance, challenges related to energy consumption, mobility management, and security must be addressed. Future work should focus on AI-based optimizations and blockchain-based security to ensure the long-term sustainability of 5G-enabled MANETs.
Video text retrieval is a hot research topic in artificial intelligence, with the core challenge being the semantic gap between visual dynamic features and discrete linguistic symbols. In recent years, with the development of large-scale models, cross-modal modeling capabilities have significantly improved, driving continuous evolution in retrieval methods regarding granularity modeling strategies. This article provides a systematic review of research methods in video text retrieval, categorizing them into single-granularity retrieval and multi-granularity retrieval. Single-granularity retrieval focuses on modeling a single semantic layer. Coarse-grained methods achieve efficient retrieval through global feature matching using pre-trained models, but they suffer from incomplete semantic coverage. Fine-grained methods enhance semantic analysis accuracy through local alignment mechanisms, but they are constrained by inherent limitations. In contrast, multi-granularity retrieval combines global scene understanding and local detail perception through hierarchical feature fusion strategies, with typical technical approaches including dynamic fusion frameworks. Analysis results indicate that multi-granularity retrieval can more comprehensively capture cross-modal semantic associations, providing a more effective solution for video text retrieval.
With the deep integration of mobile Internet and big data technology, mobile Social networks (MSNs) have become an indispensable digital living space in modern society. However, the real-time collection, analysis and sharing of massive user data leads to a systematic and complex trend of privacy leakage risks. This paper analyzes new challenges such as predictive deanonymization attacks (PDAA), lack of context awareness, and quantum computing threats, and proposes an innovative multilayer dynamic privacy protection architecture (MDPPA). By integrating key technologies such as differential privacy optimization mechanism, federated learning system, blockchain technology and artificial intelligence monitoring, the architecture constructs a context-aware, user-controllable and resilient privacy protection system. Experimental results show that the proposed model improves accuracy by 14% and communication cost by 90% compared with traditional federated learning in extreme non-independent and identically distributed data scenarios, while providing verifiable privacy protection. The research further discusses the future development direction of anti-quantum encryption and cross-chain interoperability, and provides theoretical support and technical path for building a "user-centered" next-generation privacy protection paradigm.
To gather and analyze physical data from the environment, Wireless Sensor Networks (WSNs) are created by a large number of spatially dispersed nodes on the wireless network. Problems with cluster head transmission line construction, redundancy of working nodes, and random selection of cluster heads all had an effect on the network and its energy usage. Because of this energy limitation, both the network longevity and energy-efficient routing are affected. Research on methods to reduce energy usage and increase the network's lifespan is, hence, focused on these two issues. Network performance is affected by energy consumption, communication overhead, and data transmission reliability, all of which are influenced by the fundamental process of cluster head (CH) selection. Using the Lungs Performance Optimizer (LPO), a bio-inspired optimization algorithm developed to improve the efficiency of CH selection and extend the lifetime of the network, to present an enhanced method for CH selection in this study. Our approach utilizes a fitness function incorporating key performance metrics: Average Intra-cluster Distance (AID), Average Sink Distance (ASD), Residual Energy (RE), and Cluster Head Balancing Factor (CHBF). The LPO algorithm ensures that CHs are optimally placed to minimize intra-cluster communication costs, maintain optimal distances from the sink node, and evenly distribute energy consumption among sensor nodes. Experimental evaluations demonstrate that the LPO-based CH selection method significantly outperforms conventional clustering techniques in terms of energy efficiency, network lifetime, and CH load balancing.
In this study, an integrated system combining image processing and ultrasonic sound technologies is proposed to remove unwanted animals from certain areas without causing harm. Using a deep learning-based object recognition algorithm (YOLOv11), animal species such as pigs, dogs and pigeons are detected in real-time with high accuracy. After detection, ultrasonic frequencies specific to each species - 13 kHz for pigeons, 22 kHz for dogs, 29 kHz for pigs - are transmitted to the environment to remove the animals without harming them. The system was trained with opensource datasets and images obtained from the field, and prototyped to run on Raspberry Pi 4B considering portability and energy efficiency. In field tests conducted under real-life conditions, the detection accuracy of the system was measured as 95.5% with mAP@0.5, while the removal success was above 85% on average. With its real-time response time (40 ms) and species-based frequency adjustment, the system offers an ethical, low-cost and environmentally friendly solution that can be used in agriculture, urban security and industrial areas.
The agricultural sector is undergoing a digital transformation to meet the demands of a growing global population, climate variability, and the need for sustainable resource management. The Internet of Things (IoT) has emerged as a critical enabler of smart agriculture, integrating sensor technologies, wireless communication networks, and data transmission protocols to support real-time monitoring and decision-making. This review paper provides a comprehensive analysis of IoT applications in agriculture, focusing on soil monitoring, smart irrigation, and precision farming. It examines key components such as wireless sensor networks (WSNs), data acquisition systems, communication protocols (e.g., LoRaWAN, Zigbee, NB-IoT), and cloud-based analytics platforms. The paper also discusses the benefits of IoT systems, including improved crop yields, reduced water and fertilizer usage, and enhanced operational efficiency. Furthermore, it outlines the current challenges in connectivity, scalability, interoperability, and energy efficiency. By synthesizing recent advancements and research trends, this review aims to guide future developments in IoT-enabled agricultural systems and contribute to the growing field of smart and sustainable farming practices.
The novel corona virus (COVID-2019) was encountered in Wuhan; it then spread at an alarming rate around the world and led to an ongoing pandemic. The public health, daily lives & global economy have been affected severely. Pneumonia is a life-threatening respiratory infection which affects lungs in humans and its main cause is Streptococcus pneumonia bacteria. Based on World Health Organization (WHO) statistics, pneumonia is the main cause of one out of every three deaths in India. It is very crucial to identify the positive cases in the initial stages to avoid the spread of this epidemic to a further extent and to treat the affected patients at a faster rate. Due to the non-availability of accurate automated toolkits, the requirements for auxiliary diagnostic tools have increased. The latest findings procured using techniques based on radiology imaging propose that this data contains information which is critical pertaining COVID19. To accurately detect this disease, artificial intelligence (AI) technique is integrated along with technology based on radiological imaging. This is also beneficial in identifying pneumonia and COVID-19 in the absence of experts in the medical field. This research presents an accurately designed model for COVID-19 detection with the help of X-ray’s. The model designed, classifies the diagnosis accurately into COVID-19, No-Findings and Pneumonia.
Organizations grapple with numerous cyber security challenges on daily basis. Many operate online with limited technical controls in place that cannot adequately detect, report and mitigate against the inherent risks. This paper postulates that timely and accurate detection and control of logical threats and vulnerabilities within a corporate information system’s environment is a precursor for success in implementing an effective information system security progrmame. Consequently, in this research, a five-level stratified vigilance structure is developed- and defined as: very low vigilance, low vigilance, moderate vigilance, high vigilance and hyper vigilance- then implemented in a next generation unified threat management (UTM) firewall deployed at the gateway of a high traffic corporate network to monitor activities on the network including threats and vulnerabilities. The UTM was initially deployed with the lowest level of vigilance, then gradually and systematically re-adjusted in a carefully calibrated manner, to represent the various successive levels of vigilance and handled logical threats and vulnerabilities at the various levels. The outcome of the study demonstrated that vigilance is a very important factor in detecting and controlling information systems threats and vulnerabilities in a timely and accurate manner, and this directly translates to the security of the infrastructure.
Microstrip patch antennas are fundamental to modern wireless communication, valued for their compact size, cost-effective fabrication, and adaptability across diverse applications like Wi-Fi, mobile networks, satellite communication, and IoT systems. The performance of these antennas hinges on the dimensions of the radiating patch, which directly influence critical metrics such as resonant frequency, return loss (S₁₁), bandwidth, gain, directivity, radiation efficiency, and radiation pattern. This study systematically explores how varying the size of a rectangular patch impacts these parameters, using theoretical models, equations, and simulations. Small changes in patch dimensions can significantly alter electromagnetic wave behavior, affecting signal reflection, power distribution, and radiation characteristics. By understanding these effects, engineers can design antennas optimized for specific wireless needs, balancing performance with practical constraints like size and cost. The findings provide actionable insights for improving antenna efficiency, signal strength, and frequency response, supporting the evolution of high-performance communication systems.
DL can provide cutting-edge accuracy for technologies that utilize remote sensing including unmanned aerial vehicles (UAVs), especially enhance the remote detection abilities in responding to emergencies along with catastrophe control activities. In particular, "UAVs" that have sensors for cameras are able to function online difficult accessing disaster-affected places, process images, also notify people in the event of various catastrophes like collapsed buildings, floods, or fire. The goal is to lessen the detrimental impact of disasters on the natural world and the human population as quickly as possible. DL integration introduces high processing demands that prevent the implementation of DNN for different cases where lowlatency restrictions on inference are required for making critical decisions in real-time. Consequently, our interest of this work is the efficient classification of aerial photos taken while a UAV is in flight for emergency response and surveillance purposes. Specifically, an Aerial Photos Collection for response to emergencies systems is proposed and a comparison between many current algorithms has been done. That processing is used to propose EmergencyRes, a CNN design. With lower than 1% accuracy decrease in comparison to the most advanced techniques, it can achieve up to 20x greater performance than currently available models and operate efficiently on low-power existing platforms. It can also handle multi-resolution features utilizing atrous convolutions.
Person re identification (ReID) is a crucial computer vision task aimed at matching individuals across different camera views despite variations in viewpoint, lighting, and occlusions. Even though CCTV cameras are everywhere, person reidentification remains challenging due to their inefficiency in accurately capturing faces when they are covered or blurred, as most cameras are optimized for capturing faces from a single direction. The lack of datasets that focus on multi-view capturing of individuals from all four sides makes this issue worse, with few models addressing the need to reidentify individuals based on multiple images from various angles. To tackle this problem, we have employed a novel technique that leverages the keypoints and gait of individuals from a multi-dimensional perspective, rather than relying solely on facial or biometric images. Our proposed novel SNK-PU dataset addresses this gap by including images of individuals from the four specific orientations: front, back, left, and right. Additionally, it incorporates supplementary parameters such as height and age for ReID. We trained a Siamese model with a triplet loss function on the proposed SNK-PU dataset, utilizing our innovative image generation technique. When the entire dataset is used for both training and testing, the model achieves an accuracy of 95.11%. However, with an 80:20 train-test split, the accuracy reaches 86.45%, reflecting a more realistic evaluation scenario. Unlike traditional methods that rely on uniform image recognition, our approach leverages easily accessible keypoints, enabling more consistent and robust person re-identification across diverse conditions.
Advancements in neuroscience, cybernetics, and psychology have catalyzed the nascent interdisciplinary field of neuropsychology and cybersecurity(neurocyberpsychology). This emerging area of research investigates the intricate interactions between the human brain, cognition, and technological systems synthesizing the prime developments in the origin, growth, conceptualisation, methods, applications and future horizons. Early roots lie in the cybernetics as well as neural, cognitive, and social psychological sciences provided historical groundwork for present-day neurocyberpsychology. Core conceptual frameworks integrate theories of neural computation, extended cognition, neuroplasticity, and human-computer interaction. Neuro-imaging, data mining, brain-computer interfacing, and virtual reality techniques comprise main research methodologies. Major current foci include augmented cognition, brain-computer interfaces for motor and sensory function, cognitive neural modelling, virtual reality systems, and impacts of social media and video games on the brain. Clinical uses are also emerging in neural prosthetics, neurofeedback therapies, and technological aids for psychological conditions. At the same time, neuroethical considerations spotlight dilemmas regarding security, identity, normality, addiction, and human dignity within an increasingly computerized society. To advance human flourishing amidst accelerating technological immersion, the review highlights priorities around integrative knowledge, cognitive augmentation, participatory research, interdisciplinary learning, and compassionate ethics.
----------------------------------------------------------------------ABSTRACT-------------------------------------------------------------- This paper highlights usability of statistical fuzzy inference systems based on PCA for tacit knowledge modeling in Biomass energy. Tacit knowledge is the key to management of ecological innovations in electric utilities of biomass power plants. However, tacit knowledge in the process of information gathering has not been implemented in a formalized way. Although the Sugeno defuzzification method is considered to be the most computationally effective, there is uncertainty about the defuzzified output, since it generates singleton fuzzy values objectively and is not well evaluated. A methodology enhancing the ability of statistical fuzzy inference systems for the improvement of tacit knowledge modeling. This has been exploited in the process of the new approach. The statistical inference system has been used directly integrated with the principal component analyzer, fuzzy inference engine, knowledge base and user interface. The system has been evaluated by a sub field of power systems domain of electricity marketing in biomass power plants. Approach evaluates performance of Biomass energy available for electricity generation. The has been built based on the proposed model and tacit data sets in electricity generation to be used to verify the accuracy of the model and the prototype. The electric utility assessment tool based on a questionnaire to classify electric utilities (agricultural residues, Forest waste, and Household waste) in percentages and identify electric utility performance index in biomass power plants. The experiments were conducted to investigate performance of PCA based Statistical Fuzzy Inference System based on a pilot survey. Output of the system is for computing biomass performance index based on fuzzy output values (agricultural residues, forest waste, household waste) has been generated by PCA based defuzzification process. The accuracy of the PCA based Statistical Fuzzy Inference System approach is 97%.
Support Vector Machine (SVM) is a widely used supervised learning algorithm known for its robustness, efficiency, and theoretical reliability. This study analyzes the performance of linear kernel SVMs on real-world binary classification tasks from healthcare and social behavior domains. Two datasets were used: a heart disease dataset with 14 clinical features, and a social network advertisement dataset with demographic features like age and salary. Both were pre-processed using normalization, label encoding, and train-test splitting, and implemented in Python using scikit-learn. Evaluation metrics included accuracy, precision, recall, F1-score, and confusion matrix. The heart disease model achieved 83.3% accuracy, while the social network model reached 76.7%, showing linear SVM’s effectiveness on structured, moderately sized datasets. The study also covers core SVM concepts like support vectors, hyperplanes, and margin maximization. Soft margin classification was used to handle noisy data, and hyperparameter tuning (e.g., the regularization parameter C) improved performance. In conclusion, linear SVM is still a strong baseline for binary classification due to its simplicity, speed, and interpretability. Future work may explore multi-class problems, kernel comparisons, and ensemble methods to expand its applicability.
The evolution of educational technologies has driven the need for personalized and scalable platforms in web development training. This study presents the integration of GrapesJS, a visual web builder, with Amazon Web Services (AWS), enhanced by AI-powered personalization through AWS SageMaker. The goal is to deliver adaptive, engaging, and efficient learning experiences. The paper details the system architecture, implementation strategy, and comparative analysis of student engagement and learning outcomes in AI-personalized versus traditional environments. Findings show that the AI-enhanced platform significantly improves learner motivation, task completion rates, and comprehension, thereby validating the approach for broader educational deployment.
Face detection and recognition are critical tasks in computer vision with applications in security systems, biometric authentication, and human-computer interaction. This paper presents a comprehensive study leveraging Principal Component Analysis (PCA) and Eigenfaces for efficient dimensionality reduction and compact, discriminative facial feature representation. The study introduces a robust pipeline integrating preprocessing, feature extraction, and efficient training. Using the CelebA dataset for training and the LFW dataset for evaluation, the system addresses real-world challenges, including variations in lighting, expressions, and poses. The performance is analyzed across configurations, exploring the tradeoff between dimensionality reduction and recognition accuracy. Experimental results demonstrate that the PCA-based approach achieves high recognition accuracy while maintaining computational efficiency, making it suitable for resourceconstrained environments. The findings highlight the system’s robustness, scalability, and practical applicability in both constrained and real-time scenarios. This work concludes with an analysis of strengths and limitations and offers recommendations for integrating non-linear techniques and advanced learning models to further enhance scalability, accuracy, and real-world performance.
Handling missing data is a critical challenge in data processing, as it can significantly impact the performance of machine learning models. This paper proposes a Correlation Assisted Support Vector Machine (CA-SVM) based imputation method that integrates correlation analysis with the predictive power of Support Vector Machines to enhance missing data prediction accuracy. The CA-SVM approach identifies highly correlated attributes to guide the SVM in estimating missing values more effectively, preserving the underlying structure of the dataset. A comparative study is conducted between the proposed CA-SVM, standard SVM, and K-Nearest Neighbors (KNN) imputation techniques using benchmark datasets. Experimental results demonstrate that the CA-SVM method achieves a prediction accuracy of approximately 97 per cent, outperforming both traditional SVM and KNN methods in terms of accuracy, robustness, and consistency. The findings confirm that correlation-based feature guidance significantly improves the imputation performance of SVM models. This research highlights the potential of hybrid imputation techniques in building more reliable and accurate data-driven systems.
In this paper we describe an automatic pronunciation error detection and feedback generation system for non-native second language learners by using deep acoustic model based on factorized TDNN and language model with phoneme error model. Our system builds language model considering phoneme error patterns and gives a useful feedback for learners. Deep acoustic model consists of TDNN-F with grouped fully-connected layers and shuffle operation. This network architecture maintains recognition accuracy like traditional TDNN and costs less then it. Also, our system evaluates pronunciation proficiency of utterance in word level and phoneme level based on confidence from Minimum Bayesian Risk decoder, feedback is generated on phone error model of L2 learners. This system based on resource-efficient deep acoustic architecture can be deployed in resource-limited mobile devices.
Flying ad hoc networks formed by coordinated unmanned aerial vehicles are a critical emerging paradigm for modern wireless communication in civilian, industrial, and military sectors. As UAVs become more mobile and intelligent, ensuring seamless handover in such 3D dynamic environments presents significant challenges. This paper surveys recent advancements and core challenges in handover management in FANETs, focusing on mobility issues, routing complexities, and emerging technologies including machine learning and optical communications. Moreover, a novel energy efficient handover management approach is proposed in this work leveraging dynamic operational modes such as active, sleep, and hibernate states for UAVs supported by reinforced learning techniques. The proposed framework lays the groundwork for future implementation and evaluation of energy-aware UAV operations, addressing key challenges in prolonging UAV mission duration while maintaining seamless connectivity.
With the exponential growth of mobile devices and the increasing prevalence of malicious software (malware) targeting Android devices, the need for effective antivirus solutions has become crucial. This research paper presents an analysis of the performance and effectiveness of antivirus applications (Bitdefender,Norton,Kaspersky,McAfee,Avast and trend micro) in detecting and stopping malware on Android devices. Through a comprehensive empirical study, we evaluate the top five antivirus apps available on the Google Play Store, considering factors such as detection rates, false positive rates, resource consumption, and scanning efficiency.Bitdefender was the best performer showing an 84% malware test detection rate. The research showed a good disparity in detection among top malware programs and aid users and developers in making informed decisions regarding malware protection on Android devices.
