Edge Computing: Empowering Real-Time Decision-Making K. Ulagapriya, Cynthia Jayapal, K. V. M. Shree, A. Poonguzhali, Jad Gergi Matta, Divya Bairavi Intelligent Mobile and Iot Ecosystems Bridging Cloud Fog Edge and AI, 2026 This chapter explores the transformative role of edge computing in enabling real-time decision-making across diverse domains, such as autonomous systems, healthcare, industrial automation, and smart cities. By decentralizing data processing and bringing computation closer to the source, edge computing significantly reduces latency, optimizes bandwidth, and enhances data privacy. The chapter details architectural models – including device-to-cloud, fog computing, and hierarchical edge frameworks – and highlights the integration of AI for edge intelligence. It also examines challenges like resource constraints, security vulnerabilities, and interoperability, offering insights into emerging solutions, such as federated learning, lightweight cryptography, and blockchain-based authentication. Through real-world use cases and future trends, the chapter underscores edge computing’s potential to revolutionize mobile ecosystems, energy efficiency, and secure data handling in distributed environments.
Quantum-Driven Vector Fusion Networks for Accelerating Early Cancer Detection with Machine Learning Ahila. A, P.Hosanna Princye, S.Ahamed Nishath, A Poonguzhali, Prakash R. V 2025 International Conference on Metaverse and Current Trends in Computing Icmctc 2025, 2025 Patient survival rates can be increased through early cancer detection but high-dimensional data feature redundancy and predictive inefficiency are issues with traditional diagnostic techniques. A new approach to improving feature extraction and selection and drug resistance prediction is quantum machine learning (QML) which combines machine learning and quantum computing. A novel TM with improved algorithms for early cancer detection is presented in this study. Prior to starting any experiments, the Quantum-Normalized Adaptive Refinement (Q-NAR) method was used to retrieve and preprocess the data in order to improve data understanding. Next WCAA provides the best feature representation for classification by ranking significant biomarkers. The intelligent selection of several pertinent features using Swing L-Bee Mustard Optimization (SLBMO) lowers dimensionality while increasing diagnostic precision. The last method is the Quantum Boosted Vector Fusion Network (QBVFN) which combines machine learning and quantum learning to predict cancer and analyse treatment outcomes. The suggested methodology is evaluated using the Cancer Genome Atlas (TCGA) dataset in a Python environment to show how it may enhance early-stage cancer detection feature optimization and individualized treatment plans. This study establishes the foundation for next-generation cancer diagnostics that use quantum methods in the preprocessing and predictive modelling stages which will increase clinical applicability reduce computational costs and improve accuracy
A machine learning predictive framework for diabetes management using blood parameters A Poonguzhali, P Ramkumar, Reji Thomas, S Tamil Selvan, Angel Latha Mary Artificial Intelligence in Medicine, 2024 The global prevalence of diabetes mellitus, commonly referred to as diabetes, underscores its significance as a widespread metabolic ailment. Diabetes stands as a consistent contributor to mortality worldwide, warranting substantial attention. The World Health Organization (WHO) estimates that the current number of individuals grappling with this condition reaches approximately 415 million. Projections based on comprehensive assessments indicate that our nation is poised to host nearly one million Type II diabetes patients by 2030. The implications of diabetes mellitus are far-reaching, encompassing various complications such as cardiovascular diseases, renal disorders, neuropathies, and diabetic foot issues. Addressing diabetes through early prediction and effective management holds the potential to yield enhanced patient outcomes and significant reductions in healthcare expenditures. This chapter introduces an innovative predictive framework centered on key blood parameters, including HbA1C, serum cholesterol, and serum albumin, among other variables, to anticipate the onset of diabetes. Leveraging the K-Nearest Neighbors (KNN) classifier, an analysis was performed on data obtained from 140 patients, yielding an accuracy rate of 71.42%. These findings posit that this approach can serve as a valuable tool for healthcare professionals, facilitating improved diagnosis and management of diabetes. This study underscores the potential of the proposed methodology in enhancing the healthcare landscape pertaining to diabetes.
Artificial Intelligence For Realtime Face Recognition Attendance Using College Classrooms and Buses Ahila A, Hosanna Princye P, Poonguzhali A, Kavivendhan, M. Deepa, A. Arthy, R. Saravanakumar 2024 International Conference on Electrical Electronics and Computing Technologies Iceect 2024, 2024 Teachers take attendance by having pupils sign in or check-in classes and transportation. Student absences often result from individual mistakes. This article examines a technology that records data from classroom photographs of every student's face. This research uses an Adaptive Boost Classifier, Random Forest (RF), and Deep Convolutional Neural Networks (DCNNs). The model performs well on the DCNN model with 88 and 92% accuracy and on the ResNet50 pre-trained model with 97.21% accuracy. After detecting each student's face, they recorded their present status in an Excel document. It kept the best system implementation approach based on performance.
Map Building of Indoor Environment with Sensors using Neural Network S Angel Latha Mary, K. Ulagapriya, A Poonguzhali, R. Menaha, Beaulah David, T.R. Priyadharshini Winter Summit on Smart Computing and Networks Wisscon 2023, 2023 The necessity of a blueprint of a building structure is a mandatory requirement for any reconnaissance or rescue operations. In our project, we build a modular system combining sensors related to sonar, laser, micro-wave to read sensory values and generate a 2D path of any building. The data is fetched and stored to feed to anOptimal Neural Network (ONN)-based computing system to create a 2D route with minimal discrepancies of error. Here the NN architecture is fine-tuned using Modified Dolphin Partner Optimization (MDPO) Exploration of unknown environments and space using autonomous vehicles has recently gained good attention in the field of Robotic Mapping. The recent advancements in the Internet of Things have enabled us to establish an ideal frame of reference for sonar and lidar-based systems. New effects are displayed by the sensors according to the physical characteristics of a room. The range data from sensors in various surroundings are interpreted by NNs. The distorted errors due to the material medium, particles, and moving objects present in the environment pose a threat to building a high-quality path map. The sensor fusion technique is applied to the rotatable modular array sensor to minimalize discrepancies caused by cloth materials during sonar readings, particle noise in an environment for Lidar reading, and moving human bodies present in the environment for path building and obstacle detection.
Authorization Method of Control in Android Application Using Adminio with Context-Based Access Devices A. Poonguzhali, G. Premalatha, A. Abinaya, R. Thiagarajan, R. Krishnamoorthy, S. Arun 8th International Conference on Smart Structures and Systems Icsss 2022, 2022 Students could be permitted in using smart phones with context-based handset policy initiatives as they can deactivate those application areas using the camera as well as any handset assets and permissions which students constrain while being in class, whereas the user device could indeed retain all of its original permissions outside of the workplace. Once an app has indeed been granted the required permissions upon deployment, an Android user has no command over its functionality. So, by using cbac mechanism, a new prototype has been initiated on Android, namely Adminio. Adminio app features the control over the access mechanism of the user within gps location. It tracks down the user apps and the admin can able to restrict the user to use the mobile apps within the location. By using certain privileged policies, the admin controls the user such as blocking it. Even after uninstallation, the detach IP can be identified.
INDUSTRY EXPERIENCE
One year as a production executive in a PCB manufacturing company.
