Vijaya Gunturu
@sru.edu.in
Professor ECE Director Extension Services (formerly: Dean SoE & Dean R&D)
SR University
RESEARCH, TEACHING, or OTHER INTERESTS
Biomedical Engineering, Signal Processing, Electrical and Electronic Engineering, Multidisciplinary
Scopus Publications
Scopus Publications
Vijaya Gunturu, Nekkanti Renu, Navdeep Dhaliwal, Ippa Sumalatha, Anandhi R J, and Gourab Dutta
IEEE
The number of Internet of Things (IoT) devices has grown dramatically with the technology's rapid development. Higher security standards have so been proposed for the administration, transfer, and archiving of vast amounts of IoT data. But security problems like data theft and forgeries are likely to happen while IoT data is being transmitted. Furthermore, the majority of data storage options now in use are centralized, meaning that a centralized server handles both data maintenance and storage. The confidentiality of IoT data would be seriously jeopardized once a hostile assault targets the server. Given the aforementioned security concerns, a secure transmission as well as storage solution for blockchain sensing images in the Internet of Things is put forth. Therefore, to enable effective secure data storage in Internet of Things-related smart computing systems, develop and build a novel blockchain-based artificial intelligence model. We also demonstrated the operation of the system framework. Upon conducting a thorough security study, we have determined that our suggested solution possesses a strong potential to address the majority of security issues that conventional systems encounter. Furthermore, our suggested method can be used for any file-changing wireless Internet of things network that requires the exchange of multimedia data, including traffic data from smart cities, wearable device data, healthcare data, etc.
Vijaya Gunturu, Yadavalli Devi Priya, Gayatri Vijayendra Bachhav, K Praveena, Anandhi R J, and Navdeep Dhaliwal
IEEE
Large neural language model like BERT can be pre trained to get extraordinary profits through multiple natural language processing task. Though, General Domain Corpora including web and news wire are focused on pre training efforts. The main specific pre training are benefited from general domain language models is considered as a prevailing assumption. The study focusses on the domain specific language model with abundance of unlabeled text like biomedical natural language processing and pre training from its scratch that results in more gains over the general domain language model. The investigation can be facilitated by compiling of biomedical NLP data sets that are publicly available. The experiment shows the pre training of domain specific model that act as a solid foundation in performing biomedical NLP task in wide range. the model is evaluated for modelling choices including task specific fine tuning and pre training. BERT models have some common practises involving named entity recognition using complex tagging schemes. The research can be accelerated with biomedical NLP for pre training and task specific model for the biomedical community and the leader board is created for biomedical language understanding and reasoning benchmark (BLURB).
Vijaya Gunturu, Niladri Maiti, Babacar Toure, Pankaj Kunekar, Shaik Balkhis Banu, and D Sahaya Lenin
IEEE
Transfer learning has emerged as a highly effective method for classifying biomedical images, as it entails the use of pre-trained neural networks on large and diverse datasets. The efficacy of models is substantially enhanced by this method. The challenges associated with training deep learning models from inception, such as the limited availability of annotated data and high computational costs, are circumvented by this strategy. Transfer learning expedites the training process by enhancing the precision and applicability of pre-trained models through the use of domain-specific biological imagery. By examining the various applications and techniques of transfer learning in biomedical imaging, this study investigates its potential future, drawbacks, and benefits.
Mandala Bhuvana, Kesireddy Rajashekar Reddy, Varagani Ramu, Bochu Sai Vardhan, Vijaya Gunturu, and Delukshi Shanmugarajah
CRC Press
Vijaya Gunturu, Charanjeet Singh, Nikhil S. Patankar, and S. Praveena
De Gruyter
Vijaya Gunturu, R. Krishnamoorthy, M. Amina Begum, R. Jayakarthik, Kazuaki Tanaka, and Janjhyam Venkata Naga Ramesh
CRC Press
Mandala Bhuvana Reddy, Rajashekar Reddy, Varagani Ramu, Bochu Vardhan, and Vijaya Gunturu
IEEE
Drones have emerged as a promising solution to deliver medicines and healthcare supplies to remote and inaccessible areas. This research study focuses on the use of drones to supply medicines to remote areas. The paper discusses the benefits of using drones, including their ability to reach areas with poor road infrastructure, reduce delivery times, and improve healthcare access for underserved communities. Also, this study analyses the challenges in implementing drone delivery systems, such as regulatory barriers, technical limitations, and public perception. Finally, case studies of successful drone delivery programs for medical supplies are presented and the potential for scaling up these initiatives in the future are discussed. Overall, this study argues that drones have the potential to revolutionize the delivery of medicines and healthcare supplies to remote areas and that further research and investment in this area are necessary to fully realize their potential.
Nada Tahani, Shayaan Hussain, Kunta Nithya Sri, and Vijaya Gunturu
IEEE
Automated detection in medical diagnostics such as imaging has become an emergent field. There is always a risk of exposing patients to issues brought on by human error in all areas of medicine. Among the most promising Machine Learning (ML) applications in the medical field is the analysis of chest X-rays (CXR). Yet, because of the complex structure of radiographs, the accurate identification and classification of specific diseases in CXR collections is still a challenging problem. A study reported that the level of the range of inaccuracy in radiology differs based on the medical examinations and is between 2 and 30%. The likelihood of errors also increases during the night shift hours, whereas a lower chance during the day shift hours. These mistakes may have devastating consequences for falsely diagnosed patients. This underlines the necessity for automated abnormality identification in x-rays, which might reduce the possibility of mistakes and produce reliable findings. This research study intends to enhance a medical instrument with good accuracy and specificity by assessing the results of different ML techniques.
Vijaya Gunturu, Jarabala Ranga, C Ravindra Murthy, B. Swapna, Allam Balaram, and Ch. Raja
IEEE
Artificial intelligence and fifth-generation mobile networks (5G) are two of the few technologies that work together so effectively. The complexity of a 5G ecosystem necessitates a level of service assurance that can't be met by human effort alone. Fifth-generation network operators will leverage AI for network diagnostics, cybersecurity, and personalized applications, all of which will radically alter the relationship between businesses and their consumers. The facts that even more nearly half of internet services have already integrated another aspect of AI into its 5G networks demonstrates that the industry has already invested much in the coupling. At the core of the complementary connection between AI and 5G is the significance of data; 5G opens a floodgate for data, something AI could then analyses as well as learn from more swiftly to create unique consumer experiences that were already suited to the numerous needs of consumers. As AI runs simulations for analysis, reasoning, data fitting, clustering, including optimization, the 5G network provides support in the background, increasing the reliability and significance of the findings. The eventual expansion of 5G-powered Internet of Things (IoT) devices in settings as diverse as manufacturing floors and driverless cars will cause the volume of data to grow not linearly but exponentially. Reason being, 5G will provide for much quicker data transfers. All of these gadgets and their integrated sensors provide new data that can be used to train smarter computer systems.
Vijaya Gunturu, Vipul Bansal, Manoj Sathe, Ajay Kumar, Anita Gehlot, and Bhasker Pant
IEEE
Electronic gadgets now work closely together as a result of the growth of the Internet of Things (IoT) and Mobile Edge Computing (MEC). Strong dependability and integrity of the technologies are required. But the present trust procedures have the big drawback: (1) they significantly depend on a trustworthy third party, which might have serious security problems if it were compromised; and (2) they subject the associated devices to malicious assessments that may skew their trustrank. In this research, we propose a blockchain-based done to assure for global IoT devices that incorporates the ideas of financial planning and cryptography. The link prediction mechanism uses norms trust but also risk metrics to quantify trustrank, and a novel storage structure is created to make it easier for the domain secretariat to spot and remove fraudulent assessments of the objects. Evidence suggests that the suggested trust system can protect IoT devices from malicious assaults while also guaranteeing data exchange and consistency.
Vijaya Gunturu, P. Rajani Kumari, S. M. Chithra, Bhargabjyoti Saikia, Rajesh Singh, and Devesh Pratap Singh
IEEE
The present article discusses the use of stream processing to gather data from large-scale WIFI networks. Along with the foundational techniques for deliberate sampling, data collecting, likewise network monitoring in wireless networks, we also examine how understanding extraction may be viewed as an ML problem for applications for large-scale data streaming. We highlight the major This article discusses advancements in large data stream processing methods. We also look more closely at the database collection, edge detection, and methods for machine learning that may be used in the context of WIFI analytics. We discuss challenges, academic research, and the results of wireless network monitoring and stream analysis. Further research is anticipated into other dataflow improvements, such as pattern recognition and optimization algorithms.
Vijaya Gunturu, V. Savithri Padma Priya, Ebenezer Abishek B, Shreya Shetty, Dharam Buddhi, and Surendra Kumar Shukla
IEEE
Corporate IT architecture benefits greatly from cloud services, which is supported by virtual machine, which makes resource centralization simple and reduces cost, space, and administrative work. Nevertheless, a lot of clients are still hesitant to Due to worries about the security of sensitive and confidential data, companies are making the shift from their traditional internal IT infrastructure to a cloud storage service. The possibility of corporate data breaches and privacy losses is increased by the methods for physically locating the popularity of vulnerability assessment exploits, cloud-based virtual machine instances, and bridge assaults across virtual servers. This research offers a strategy for reducing such risks and fostering client confidence in business cloud computing. Even with well designed software systems, new flaws are found on a daily basis, and hacking methods are evolving over time. This situation makes it seem unlikely that a corporation computer processing will be completelysecure; software companies in the internet are prone to security breaches. A well-designed attack contingency plan is the practical answer to the security issues. On the plus side, if an adequate defensive mechanism is in place, cloud computing's communal infrastructure may be exploited to successfully neutralize threats. We suggest such a cloud-based threat mitigation architecture. Various software flaws in the cloud have varying degrees of severity and effects on security measures (confidentiality, integrity, and availability). We continually track and assess the risk of penetration in various security variables using the Markov model (For instance: a change in the potential to undermine the confidentiality of the data.) Our methodology would make it easier for tenants to determine their ability to deploy a dynamic mitigation approach if there is a Meanwhile to Security Flaw (MTTSF) a drastic shift in risk. This architecture adds an additional layer of safety to thedifferent cloud management, which can increase consumer confidence in business cloud services.
H. Simitci, C. Malakapalli, and V. Gunturu
IEEE
This study explores the performance implications of an IP storage protocol (iSCSI) in a storage network. The results are compared with the characteristics of the fibre channel protocol. We set up a test-bed consisting of Linux PCs connected together with both fibre channel and Gigabit Ethernet adapters. We conducted experiments on user and kernel level TCP/IP communication and iSCSI data transfers over TCP/IP. By instrumenting the network and iSCSI device drivers on the initiator and target machines, we have collected performance data on various aspects of the protocols. Using a prototype SCSI target mode driver on the target machine, we were able to conduct similar experiments with the Fibre Channel protocol and interconnects. We present preliminary performance data for iSCSI and ways to improve the underlying TCP/IP bandwidth on Gigabit Ethernet.