A novel enhanced security architecture for sixth generation (6G) cellular networks using authentication and acknowledgement (AA) approach Samuthira Pandi V, Anitha Juliette Albert, K. Naresh Kumar Thapa, R. Krishnaprasanna Results in Engineering, 2024 The sixth generation (6G) of cellular transmission is extremely adaptable. 6G will equip everyone with pervasive wireless connectivity and meet the needs of a fully connected world. It is anticipated that revolutionary ideas will increase in support for a fast-expanding variety of smart devices and applications. The sophisticated characteristics of 5G portable cellular network infrastructures produce novel threats and requirements. Comparing 6G cellular network technologies to 5G cellular networks, this paper discusses various security and privacy issues in 6G cellular networks based numerous security services. Proposes AA approach to evaluate the effectiveness of the methodology. Propose a new 6G wireless security architecture based on the analysis of secret key authentication and flexible position based identification, which serves as the foundation for an examination of identity management and flexible authentication. Demonstrates the benefits of the proposed architecture, analyzes BER against SINR and Measure Throughput Against various SINR values. Finally, Limitations and future recommendation for 6G cellular network security is presented.
Advancing Pneumonia Diagnosis: Hybrid and Optimal Deep CNN Model for Chest Image Classification Gunapati. Suresh, T. Ravi, R. Krishnaprasanna 2nd IEEE International Conference on Integrated Intelligence and Communication Systems Iciics 2024, 2024 Pneumonia is a significant worldwide health issue, with numerous etiologies, such as bacteria, fungi, and viruses, contributing to its challenging diagnosis. Computed Tomography (CT) imaging is crucial for diagnosing pneumonia because it provides detailed information on lung abnormalities. Many hospitals have adopted CT scans during the pandemic to identify lung diseases caused by respiratory infections. Not only is it more expensive, but it is not as widely accessible. Early, precise, and cost-effective identification of pneumonia problems is necessary for improved treatment outcomes. Pneumonia-Plus, which uses Convolutional Neural Network (CNN) architecture, was trained on a large set of annotated CT scans that included bacterial, fungal, and viral pneumonia cases. This work aims to optimize the system to recognize distinct features associated with multiple independent causes of pneumonia. This work utilized a Deep Learning (DL) model, dubbed Pneumonia-Plus, to categorize viral, bacterial, and fungal pneumonia using chest photos. The experiment shows that Pneumonia-Plus performs better overall by accurately classifying pneumonia cases. Most importantly, the model achieves the best accuracy and recall rates while showing resilience in distinguishing between bacterial, viral, and fungal pneumonia.
Classification of Microglial cells using Deep learning techniques P Chitra, Y. Beryl Vedha, S. Johnson Retnaraj Samuel, R. Krishnaprasanna, S Rohan, C S Nandha Kishore Proceedings 2nd International Conference on Advancement in Computation and Computer Technologies Incacct 2024, 2024 Microglia are specialized immune cells found only in the Central Nervous System (CNS), with 1.5 trillion numbers of them identified across the spinal cord and brain. These cells display various characteristics depending on the signals they receive, ranging from beneficial to harmful, as evidenced by their movement and physical changes. Numerous neurological illnesses, including Parkinson’s and Alzheimer’s disease, have been connected to changes in the inflammatory condition. Understanding the significance of microglial cells in disease progression and examining the feasibility of novel therapy methods in such situations. In this paper, a study is conducted on the microscopy image of microglial cells. Initially, the image is preprocessed using image processing algorithms, and the number of cells in the image is also automatically counted using the image processing algorithm. The preprocessed image is classified using Deep Learning Neural Network (DNN) to classify whether the cell is in a normal or activated state. The classifiers such as Resnet50, VGG16, VGG 19 and Xception are applied to the preprocessed image. The classifier’s performance is assessed using evaluation parameters. The results are compared to choose the model with the highest accuracy. For this data the accuracy of VGG16 net achieved is 96% which is higher than other model considered.
Experimental Analysis of Malware Detection and Classification System using Intelligent Deep Learning Methodology K. Indumathi, R. Krishnaprasanna, G Chamundeeswari, M. Preetha, Tamilvani K, Senthilkumar B Proceedings of the 4th IEEE International Conference on Smart Technologies in Computing Electrical and Electronics Icstcee 2023, 2023 Computer system security relies heavily on the ability to identify and eliminate malware. Signature-based approaches are widely used, although they are not able to effectively identify zero-day assaults or polymorphic infections. This is why there is a demand for detection methods based on deep learning and categorization. Computers, Smartphones, and apps are all fair game for malicious software. Malware is software that has to be watched out for since it might be a hazard for both computer users and internet-based systems. Malware was developed with the intention of stealing sensitive data from a user's computer or taking remote control of a user's device. Malware is software designed to disrupt or even destroy data on a user's computer. Malicious software can enter a computer in a number of ways, one of which occurs over a network that already has the virus installed. This project will investigate and deploy high-precision classification models using deep learning technologies from the discipline of computational intelligence to enhance the efficiency and accuracy of malware detection. Learning based Classification Scheme for Improved Malware Detection (LCSIMD) is a unique technique described in this work and utilized as the primary training model. The suggested LCSIMD framework is trained with malware visualization tools. Models with higher classification accuracy can be found by transforming malware features into images and then modifying their characteristics and input procedures; similarly, prototypes with proposed model can benefit from the use of suitable characteristics and preprocessing methods. The final step is to generate characteristics for network-generated instances, which improves the precision of small-scale training. Every one of us desire to utilize malware that impacts various devices as a teaching example, as described above. The output efficiency of the suggested technique is cross-validated with the classic Neural Network (NN) model to assess the performance of the new scheme.
A Novel Design of an Image Encryption and Decryption Scheme Using Enhanced Cybersecurity Principles Ganesamoorthy R, Pushpa G, Senthilkumar B, R. Krishnaprasanna, Vijayalakshmi. G, V. Samuthira Pandi 1st International Conference on Emerging Research in Computational Science Icercs 2023 Proceedings, 2023 Information security has risen to the top of the priority list as a direct result of the fast expansion of digital communication transmission and delivery. Protecting sensitive picture information is becoming more important than it has ever been before as the usage of images becomes increasingly common. The requirement to protect one's reputation has reached the level of an absolute necessity. The safeguarding of one's privacy has developed into an activity of critical importance. There is a wide range of different approaches that have been developed as a means of protecting sensitive data as well as the privacy of individuals. Images are encrypted to protect sensitive data from being viewed by unauthorized parties. Encryption is a popular practice that is utilized for the purpose of protecting confidential information from prying eyes. In this study, we enhance the performance of image encryption by utilizing a revised version of the Enhanced Image Cryptography Principle (EICP) that makes use of a key stream generator. This helps us get better results. All cryptographic approaches for sending pictures over an unsecured network make use of image cryptography approaches. The researchers have developed a number of such methods for safely and effectively communicating information over an unsafe network and these approaches are used by all cryptographic approaches. This study aims to demonstrate the restricted encryption methods that were applied to disguise the image over a susceptible network in order to accomplish its purpose. This research proposes a novel encryption method that has been given the acronym EICP. The goal of this method is to improve image cryptography in a timely manner. The proposed approach is cross-validated with the conventional cryptographic algorithm called Rivest, Shamir, and Adleman (RSA) to prove the efficiency of the proposed approach in fine manner.
A Real-Time Spam Identification Scheme over Social Networking Environment Using Deep Learning Principles Mahalakshmi P, V. Mahalakshmi, E.S. Vinothkumar, Senthilkumar B, M Dinesh, R. Krishnaprasanna 3rd IEEE International Conference on Mobile Networks and Wireless Communications Icmnwc 2023, 2023 Every single day, millions of people all around the world utilize various social networking platforms. The usage of social media platforms like Twitter and Face Book may result in both positive and bad results for users. These outcomes are not mutually exclusive. The most popular social networking sites have recently been a primary target for spammers who wish to disseminate large numbers of undesired and maybe harmful information. For example, Twitter has quickly become one of the most commonly used platforms ever, which has led to a large increase in the amount of spam that is posted on the network. Users who really use Twitter are irritated and frustrated by tweets from bogus accounts that promote meaningless products, services, or websites. There has also been an increase in the ease with which hazardous chemicals may be transmitted by presenting users with misleading information while using bogus identities. This has led to an increase in the likelihood that people would be harmed. In the field of contemporary social media, two of the most popular research subjects are the elimination of spam and the verification of users on Twitter. In the event that they continue to disseminate dangerous advertisements, spam accounts on social networking websites pose a substantial threat to the safety of the internet and should be eliminated as quickly as possible. This article explores the origins of spam accounts on social networks like Twitter, as well as the distinguishing characteristics of spam accounts, with the goal of improving spam identification. Twitter is one of the most popular online social networks (OSNs), and its members include ministers, business moguls, Hollywood actors, and Fortune 500 companies. The platform's 313 million monthly active users are responsible for publishing around 500 million tweets each and every month on it. Due to Twitter's rising popularity, spammers have developed an interest in the platform. These malicious actors exploit the service for their own nefarious objectives, like as spying on normal users, distributing hazardous malware, and promoting their own websites via links posted in tweets. They also use the service to advertise their own websites via links posted in tweets. They resort to strategies such as covertly following and not following legitimate individuals for the purpose of gathering confidential information, which is their aim. In order to solve this problem, we have developed a technique for detecting spam that is based on deep learning and has been given the name Learning Principle for Spam Identification (LPSIM). This correct proof of the efficacy of the recommended scheme is shown in the form of a histogram, and it is assessed in comparison to the conventional learning scheme that is known as an Artificial Neural Network (ANN).
