I, Dr. Nikhila Kathirisetty am working in Computer Science and Engineering (Data Science) at Vardhaman College of Engineering, Hyderabad. I have been teaching in India and the UK for more than fourteen years. At Marwadi University in Rajkot, Gujarat, I finished my doctorate in Computer Engineering in 2024. My B. Tech. in CSIT under JNTU (2005) in Andhra Pradesh, India, and my master's degree M. Tech. in CSE from Hyderabad, Telangana, under JNTUH (2010). In addition to being a senior member of IEEE, I am in charge head of the Department of CSE (Data Science) at VCE, Hyderabad, Telangana. I published papers in reputable journals including IEEE Access, international conferences, and IEEE conferences. Software engineering and machine learning are among the research and interest areas.
EDUCATION
Ph. D - Computer Engineering- 27 April, 2024
M. Tech - CSE- Jan 2010
B. Tech - CSIT - April 2005
User-Centric Internet Performance Monitoring in Smart Homes Dr. Manoranjan Parhi, Simran Raj, Dr. Gopalakrishna V. Gaonkar, Dr. Kethirisetty Nikhila, Charu, Ankit Punia Journal of Internet Services and Information Security, 2025 As smart home devices and applications become more common, maintaining strong and dependable internet performance enables their seamless use. However, the conventional approach to network performance monitoring does not capture the user experience in real time, which leads to user dissatisfaction and inefficient detection of performance issues. This paper describes an internet performance monitoring system tailored for smart homes that is focused on user experience. It implements a mixture of active and passive measurement techniques with behavioral context modeling to strategically deploy light sensor agents that gather fine-grained QoS and QoE data, including but not limited to latency, jitter, packet loss, throughput, and user ratings over time and across multiple applications. The system also uses machine learning to track network performance metrics with user satisfaction to enable realtime performance assessment and automated adaptive optimization recommendations. Tested in simulated and real smart homes, the system demonstrates improved detection of micro-disruption, improved device-specific problem detection, and improved estimation of quality of experience. The paper illustrates the context-aware evaluation of internet performance. It emphasizes smart home networks designed to be more adaptive and user-centered, responding not only to technical measures but also to user evaluation. The model suggested here boosts performance transparency for end-users and allows service providers to deliver proactive and tailored support.
Implementing Digital Twins for Real-Time Library Operations Monitoring Ankit Punia, C. J. Geetha, Dola Babu Ramesh, Kethirisetty Nikhila, Honganur Raju Manjunath, R. K Tripathi Indian Journal of Information Sources and Services, 2025 Increased user expectations for intelligent systems fuel enhanced digitization of daily activities in library systems. We propose a new framework RT-LTM, Real-Time Library Twin Monitoring, which aims to utilize Digital Twins for real-time monitoring, predictive maintenance, and tailored service optimization in library spaces. RT-LTM framework designs a virtual model of library assets such as books, RFID-tagged materials, HVAC systems, and previous behavior analytics to achieve real-time synchronization of physical and virtual systems. The architecture integrates IoT sensors, edge-cloud computing, and AI analytics for real-time monitoring of spatial efficiency, resource consumption, and maintenance forecasting. A university library case study demonstrates the active library's real-time monitoring efficacy, resource access, energy use, and user satisfaction. The results highlight the outstanding impact of Digital Twin technology in contemporary library systems and services management.
Performance improvement of image mosaicking using image inpainting Shivangi Patel, R. Rajasekar, Mohd Sadim, Nikhila Kathirisetty Digital Transformation and Sustainability of Business, 2025 A mosaic is an image created by putting together fragments of tile to create a huge perspective. Image mosaicking is a technique that allows you to see a picture from multiple perspectives more clearly by rearranging a collection of distinct or overlapped sub-frames to create a whole view. The technique of rebuilding the missing area to alter the damaged piece so that the inpainted region is invisible to users who are unfamiliar with the original image is known as image inpainting. The primary focus is on selecting the area to be inpainted in order to enhance the image quality. Finally proposed method is based on patch replacement procedure for panoramic image restoration and re-filling the image with adjacent neighboring pixels. Paper presents scale invariant feature transform (SIFT) for mosaicking and exemplar based technique for image inpainting. The parameters used: Peak signal to noise ratio (PSNR), features detected and computational time.
Skin Cancer Detection Using U-Net Ganesh B. Regulwar, Katta Hartheek Reddy, Vanam Prem Shanker, Ashish Mahalle, Venigalla Jithin, Nikhila Kathirisetty Applied Machine Learning in Healthcare Case Based Approach, 2025 Most deaths due to skin cancer are caused by malignant melanoma, which is considered one of the most dangerous forms of cancer. Malignant melanoma is curable in its early stages with a biopsy. Early detection is the most effective way to ensure a favorable prognosis for skin cancer. Medical imaging, such as dermoscopy and standard camera images, is the most suitable tool for diagnosing melanoma in its early stages. Radiologists require computer-aided diagnosis (CAD) systems to accurately diagnose melanoma. However, accurately segmenting and classifying pigmented skin lesions is challenging due to the various colors and structures that appear randomly within them. We attempted to develop a more precise and easily adaptable model for clinical use by evaluating various existing machine learning implementations against open datasets. This chapter presents a systematic approach for classifying hyperpigmented skin lesions on dermoscopy images. U-Net can be used with less training data than traditional convolutional neural networks to achieve better results. It requires less training time than other convolutional neural networkss, making it the best model for biomedical image segmentation. It is flexible and can be used for almost any image segmentation task.
Voice-Based Parkinson's Disease Stage Classification Gottumukala Gayathri, Ganesh B Regulwar, Sreeja Vanga, Arukala Pradeep, V.Muni Shekhar, Nikhila Kathirisetty 8th World Engineering Conference on Contemporary Technologies Wecon 2025 Nextgen Sustainable Technologies, 2025 The progressive neurodegenerative condition known as Parkinson’s disease (PD) impairs both cognitive and motor abilities. Predicting illness phases accurately is essential for early intervention and individualized care. This work suggests a machine learning method based on clinical and biological data to categorize and forecast the stages of Parkinson’s disease using the XGBoost algorithm. The patient records in the dataset include characteristics including voice abnormalities, motor function scores, and other pertinent biomarkers. The XGBoost model is assessed against conventional classification models like SVM. Test findings show that XGBoost outperforms other machine learning methods in terms of accuracy, sensitivity, and specificity. This strategy improves patient care and disease management by offering a dependable and effective way to forecast Parkinson’s stage. According to comprehensive comparative results, XGBoost performs better than traditional classification models like SVM in tests, with higher accuracy, sensitivity, and F1-scores throughout all Parkinson’s stage.The proposed XGBoost model achieved an accuracy of 93.02%, precision of 0.93, recall of 0.92, and an F1-score of 0.91, outperforming traditional classifiers.
Fundamentals of wireless sensor networks: Wireless sensor networks techniques S. Satheesh Kumar, Sumaiya Shaikh, Nikhila Kathirisetty, Ganesh Bhaiyya Regulwar, E. Ravi Kumar Machine Learning for Environmental Monitoring in Wireless Sensor Networks, 2024 As the current computing era is working behind digital data, it is indeed important to concentrate on climatic change, which has become dynamic. Even a few non-automation processes have been converted into automation by using IIoT and industry 4.0 revolution technologies such as data analytics, the Internet of Things, cybersecurity, and machine learning. Sensor networks (SN) play a pivotal role in the collection and transformation of data through electronic devices as sensors. The sensors that work in this stream are ultrasonic sensors that help measure the distance between two vehicles. This proposal concentrates on different sensors used in environmental monitoring, data collection, and data transformation from devices to clouds and cyberattacks that poison wireless sensor networks: tampering attacks, replication attacks, blackhole attacks, wormhole attacks, Sybil attacks, and link layer attacks. Sensors which retrograde in environmental monitoring are listed as Temperature sensor, humidity sensors, airflow sensors, pressure sensors, vibration sensors, and water measuring sensor
Audio to Sign Language Translator Ganesh B. Regulwar, Nihitha Gudipati, Mahesh Naik, Nikhila Kathirisetty, Laasya Madiga, E. Ravi Kumar 2023 2nd International Conference on Ambient Intelligence in Health Care Icaihc 2023, 2023
