School of Science & Technology (SST)

Permanent URI for this communityhttp://10.0.2.71:4000/handle/123456789/439

Browse

Search Results

Now showing 1 - 10 of 13
  • No Thumbnail Available
    Item
    Mu-eta Fusion Experimental Investigation of Wireless System Performance in Weibull Fading and Correlated Interference Scenarios
    (World Scientific, 2026) Upadhyay, Deepak; Chhabra, Gunjan
    In this paper, we have introduced the Mu-Eta fusion to compute wireless communication performance over Weibull fading and correlated interference. In contrast to typical fading models, the Weibull distribution provides adjustable parameters to represent various NLOS environments. Correlated interference, typical in high-density environments, has a severe influence on system reliability. Transmission performance analysis is presented in terms of throughput, BER, energy efficiency and secrecy capacity in an exact closed form. The robustness of the presented model is confirmed by simulation and SDR results. Results indicate that Mu-Eta can achieve significantly higher throughput than the conventional schemes and the achieved throughput can be substantially superior, with up to 36% higher throughput and 35% lower BER under high interference correlation. The framework is especially relevant to urban wireless systems, vehicular networks and the future 5G/6G generations.
  • No Thumbnail Available
    Item
    Emerging Role of Artificial Intelligence (AI) in Early Autism Diagnosis: Comprehensive Review
    (Springer, 2026) Pant, Shivani; Gehlot, Anita
    SDG 10 aims to reduce inequalities and ensure no one is left behind. Autism, a complex neurological disorder, is found in around 1/100 children worldwide. Early diagnosis is crucial for effective treatment and intervention. Artificial Intelligence (AI) has emerged as a revolutionary tool in ASD research, enhancing early detection mechanisms and improving symptom management. ML, supervised, unsupervised, semi-supervised, and reinforcement learning algorithms are used for image recognition, language understanding, and speech processing. This paper aims to emphasize the role of AI in early autism diagnosis and to identify and evaluate various AI-based algorithms used for accurate autism detection. Autism diagnosis is challenging due to the lack of standard medical tests. AI interventions, such as natural language processing, machine learning and deep learning. Can improve early diagnosis by identifying complex patterns and detecting subtle symptoms. However, issues such as algorithm transparency and data privacy must be addressed. With continued research and ethical AI technology, these algorithms can transform early ASD diagnosis, enabling timely intervention and improved outcomes.
  • No Thumbnail Available
    Item
    Enhance Reliability and Security in VANET Using Clustering Based on ANT Colony Optimization and Fuzzy Logic
    (Wiley, 2026) Khan, Gulista; Kanauzia, Rohit et al.
    Vehicular Ad Hoc Networks (VANETs) play a crucial role in intelligent transportation by enabling communication between vehicles and infrastructure. However, ensuring secure, reliable, and consistent data transfer remains challenging due to their dynamic nature. This paper proposes a Clustering-Based Ant Colony Optimization (CB-ACO) and Fuzzy Logic algorithm to address these issues. Clustering reduces network load and enhances stability, while ACO selects optimal cluster heads and routes, supported by Fuzzy Logic-based trust assessments for secure access control. Compared to existing algorithms, the proposed method improves packet delivery, reduces latency, and enhances security, offering a robust solution for next-generation VANETs.
  • No Thumbnail Available
    Item
    An Explainable 3D Graph-Transformer Framework for Hepatotoxicity Prediction with Scaffold-Aware Evaluation
    (IEEE, 2026) Mishra, Suraj; Kanauzia, Rohit et al.
    Drug-induced hepatotoxicity continues to be a major factor in post-market removal and late-stage drug attrition. While graph neural networks have shown promise in predicting chemical toxicity, many existing approaches rely on random data splits and offer limited interpretability, thereby reducing their reliability in real-world screening scenarios. This study presents an explainable 3D graph-Transformer framework for hepatotoxicity prediction that integrates geometric inductive bias, rigorous scaffold-based evaluation, and multiple complementary explainability approaches. Molecular structures are encoded, and spatial linkages are captured using a 3D-equivariant graph neural network. Next, a graph-level Transformer is used to model long-range atomic interactions. Five-fold scaffold cross-validation is used to assess the model on a curated hepatotoxicity dataset of 8,538 molecules obtained from Tox21 liver-related assays. We use three different explainability techniques to guarantee transparency: atom-level perturbation analysis, integrated gradients, and transformer attention visualization. Experimental results show that the proposed framework achieves 91.8% accuracy with perfect recall on the evaluated splits, while generating explanations that are chemically significant and consistent with recognized hepatotoxic substructures. The suggested method is appropriate for early-stage toxicity screening and decision support because it combines multi-view explainability with robust out-of-distribution evaluation.
  • No Thumbnail Available
    Item
    Future Trends: AI And Cloud Innovations In Healthcare
    (Bentham Science, 2026) Bhatt, Ashutosh; Aggarwal, Ambika et al
    Amalgamation of AI & cloud computing brings a significant relevant change in healthcare, where they are providing assistance to the doctor by examining huge amounts of data in the medical discipline. This means with the help of these emerging technologies, experts can easily diagnose the disease in the early stage, provide better personalised cure plans, speed up the treatment, and the discovery of new drugs. Cloud computing is dedicated to keeping track of data securely with accuracy while sharing it with others. But, here it is important to focus on some issues like algorithm selection, authenticity, integrity, and security of the patients’ data. Despite these concerns, AI and cloud machinery can really boost healthcare and help to create a better future with more proactive and personalised care for a healthy population. The operations of medical research, managing all aspects of patient care, and delivering the results with accuracy are evolving right now. This chapter highlights the impact of AI & cloud tech in the coming years to identify their fruitful effect in healthcare. In the bigger picture, these technologies strive to achieve remarkable changes with improvement in areas like instant drug discovery, medical imaging, virtual assistance, accurate prediction in smart health, early diagnosis, treatment, and cure. The healthcare cloud system is much more committed to delivering a structure that gives cost-effective, secure, and shareable services in healthcare along with advanced disaster recovery plans using AI tools and techniques. The real revolution with amazing possibilities can occur with the help of AI & CC (Cloud Computing) if we can provide better ways to handle population health using precise public health strategies, data processing in real-time and executing the services like AI-as-a-service (AIaaS).
  • No Thumbnail Available
    Item
    Remote Sensing Reveals Land Cover Dynamics at SIDCUL Industrial Estate, Haridwar: A Sacred City in the Himalayan Foothills, India (2018–2023)
    (Springer, 2026-04-01) Joshi, Pooja; Bhatt, Ashutosh
    This study employs Sentinel-2 satellite imagery and spectral indices (NDVI, MNDWI, NDBI) to quantify land cover changes in Haridwar, India, between 2018 and 2023. Using Google Earth Engine, we computed index differences and classified land cover into five categories: Barren, Sparse Vegetation, Dense Vegetation, Water, and Built-up. Results indicate a 5.5% expansion in built-up areas (+3.68 km2), a 2.9% decline in dense vegetation (−1.36 km2), and minor increases in water bodies (+14.1%) and barren land (+4.3%). These changes reflect urbanization pressures near critical Himalayan ecosystems, providing a baseline for sustainable land management.
  • No Thumbnail Available
    Item
    Design and Simulation of Small Scale RF MTF-MOSFET
    (Springer, 2026-05-01) Payal, Mohit; Sharma, Vibhor et al.
    In this study, trenches inside the epitaxial layer are proposed as structural alterations to the traditional planar metal-oxide semiconductor field-effect transistor (MOSFET) on silicon-on-insulator (SOI) substrates. The innovative multi-trench-finger MOSFET (MTF-MOSFET) features multiple channels in its p-base, achieved by placing several vertical gates in distinct trenches. These additional channels enable parallel conduction of the drain current. TaN is used as the gate electrode and silicon dioxide (SiO2) as the gate dielectric in the suggested MTF-MOSFET architecture. The inclusion of multiple channels significantly enhances the device’s electrical characteristics. Specifically, the transconductance (gm) and the drain current (ID) are substantially increased due to the concurrent conduction of many channels. This concurrent conduction leads to a marked improvement in the cut-off frequency (fT). Two-dimensional simulations were used to assess and contrast the MTF-MOSFET’s performance with that of a standard MOSFET. The results demonstrate that, with a gate length of 60 nm, the MTF-MOSFET exhibits: 6.5 times increase in ID, 3.7 times improvement in gm, 1.3 times enhancement in fT and superior control over short channel effects. These improvements underscore the significant advantages of the MTF-MOSFET design over the traditional planar MOSFET, particularly in terms of current handling, transconductance, frequency response, and mitigation of short channel effects. Proposed device is used for ICT Applications for Electrical and Intelligent Engineering.
  • No Thumbnail Available
    Item
    Leveraging Machine Learning in Digital Payroll and Payment Playforms Redifining HR Practices in the FinTech Era
    (IEEE, 2026) Rao, P.Venkateswara; Chhabra, Gunjan et al.
    The objective of this research paper is to explore how Machine Learning (ML) could potentially be used to transform the HR process within Digital Payroll & Payment Platforms in today's Fintech industry. We did this through development of a Reinforcement Learning (RL) model in a simulated environment, utilizing artificial data to measure various Performance Metrics including Efficiency, Fairness, Fraud Prevention and Employee Satisfaction. Our results indicated that our AI-enabled model outperformed the AI-free Baseline Model in terms of Reward Convergence, Satisfaction Index, Transaction Success Rate with Fraud Attempted, and Scalable Adaptability as the Workforce Size increased. Additionally, we conducted Ablation Testing for Features, which confirmed the importance of engineered features to produce desired outcomes. These results clearly indicate that Intelligent Payroll Platforms have the potential to reduce employee stress, increase employee trust and provide Human Resource Managers with tools to create Financial Operations that are both Sustainable and Employee-Centered.
  • No Thumbnail Available
    Item
    Quantum-Enhanced Deepfake Detection: A Hybrid Machine Learning and Quantum Computing Approach for Secure Surveillance
    (Springer, 2026-01-28) Katiyar, Vivek; Mishra, Anupama; Joshi, Bineet Kumar
    Deepfake technology poses a growing threat to cyber security by creating convincing fake images and videos. In this research paper, we propose a hybrid detection model that combines the strengths of Quantum Computing (QC) and Machine Learning (ML) to handle this issue more effectively. Our approach integrates quantum feature extraction with Convolutional Neural Networks (CNNs) to improve the accuracy of deepfake detection. The solution is tested on standard datasets like FaceForensics++ and Celeb-DF, the model outperformed traditional CNN and Transformer-based methods, achieving up to 91.8% accuracy and 91.5% F1-score. These promising results suggest that quantum-enhanced models could play a powerful role in future secure surveillance and digital forensics systems.
  • No Thumbnail Available
    Item
    A comprehensive survey on federated learning for privacy preservation in digital healthcare applications
    (Springer, 2026-01-11) Kanauzia, Rohit; Singh, Mridula et al.
    Internet of Medical Things (IoMT) is a relatively new service that has the possible to revolutionize healthcare by connecting previously analog technologies digitally. Consequently, numerous healthcare applications based on IoMT are utilized in the course of daily life. Despite the abundance of machine learning (ML) techniques aimed at improving healthcare data management, none of them have been able to guarantee the data's complete privacy and security. The precise nature of the clinical data makes ML application difficult and yields unsatisfactory results. A new paradigm in ML called federated learning (FL) has arisen as a means to discover untapped potential in digital healthcare uses that protect patients' and clients' privacy without compromising their data. This survey comprehensively reviews over 105 peer-reviewed publications (2018–2025) sourced from IEEE, Elsevier, Springer, ACM, and MDPI digital libraries. It classifies existing FL approaches for digital healthcare based on architecture, communication efficiency, privacy preservation, and application domain. Survey highlights key findings, comparative analyses with conventional FL-based systems, and lessons learned from prior studies. Finally, open challenges such as scalability, energy efficiency, model heterogeneity, and secure aggregation are deliberated, along with future research directions to enable trustworthy FL-based IoMT healthcare ecosystems.