http://dspace.dtu.ac.in:8080/jspui/handle/123456789/50 2026-04-28T05:27:44Z 2026-04-28T05:27:44Z DUBEY, MRITYUNJAY http://dspace.dtu.ac.in:8080/jspui/handle/repository/22652 2026-02-10T04:47:38Z 2022-05-01T00:00:00Z

Title: CORS VULNERABILITY TESTER FOR WEB APPLICATION Authors: DUBEY, MRITYUNJAY Abstract: The problems that were caused by the same-origin policy and the incorrect setup of it led to the development of a protocol known as cross-origin resource sharing (CORS). This protocol was designed to solve these problems. Current versions of web browsers come equipped with a feature known as the same-origin policy.. Scripts that are housed on one domain are unable to make calls to scripts that are placed on another website as a result of this functionality. This security policy may ban certain legitimate use cases that pose no security risky. Utilizing CORS is the optimal option for ensuring that those valid situations are able to work correctly.During the process of designing, implementing, and deploying CORS, we discovered a number of additional security problems, including the following: 1) CORS diminishes cross-origin "write" privilege in practical ways. 2) CORS introduces additional trust requirements the web of different interactions. 3) CORS is something which isn’t well understood for being developers, most likely as a result of its. opaque policy and complicated and complex linkages with other web protocols, which results in a variety of misconfigurations. This is the case since CORS is notoriously difficult to understand. In conclusion, we provide simplified and clarified versions of the protocol in order to solve the security problems that were uncovered by our study. Both the CORS standard and the most common web browsers have taken some of our suggestions and implemented them in a variety of different ways.

2022-05-01T00:00:00Z SINGH, HARSHIT http://dspace.dtu.ac.in:8080/jspui/handle/repository/22649 2026-02-10T04:47:14Z 2025-12-01T00:00:00Z

Title: REVIEW ON LARGE LANGUAGE MODELS Authors: SINGH, HARSHIT Abstract: Recently there has been a paradigm shift in a machine’s ability to solve various natural language processing(NLP) tasks from generating coherent sentences to in-context learning. This shift has been enabled by pre-trained transform models on large amounts of data; the research community aptly called these models as Large Language Models(LLM). In this review we will go over a brief overview ofthe technology as well as the ethical and social implications of such models and conclude with the future potential avenues for further research.

2025-12-01T00:00:00Z KHERA, SHIVANSH http://dspace.dtu.ac.in:8080/jspui/handle/repository/22644 2026-02-10T04:46:50Z 2025-05-01T00:00:00Z

Title: EMOTION-AWARE MULTIMODAL SARCASM DETECTION USING DEEP LEARNING TECHNIQUES Authors: KHERA, SHIVANSH Abstract: Sarcasm detection in digital communication has become increasingly challenging as users employ sophisticated combinations of textual and visual elements to convey ironic meaning. Traditional text-based approaches and existing multimodal methods struggle to capture the emotional incongruity that is fundamental to sarcastic expressions, repre- senting a significant gap in current multimodal sarcasm detection research. This thesis proposes an emotion-aware multimodal framework for sarcasm detection that systematically integrates emotional features from both textual and visual modal- ities. The approach enhances existing deep learning architectures, specifically Bidirec- tional Long Short-Term Memory (BiLSTM) networks and Graph Convolutional Networks (GCNs), with emotion recognition capabilities utilizing DistilRoBERTa for textual emo- tion analysis and computer vision techniques for visual emotion recognition. The frame- work was primarily developed and optimized using the MMSD2.0 dataset, followed by comprehensive cross-dataset evaluation on MMSD Original and MEMOTION datasets to assess generalization capabilities. Experimental results demonstrate that systematic integration of emotional features from both modalities significantly improves sarcasm detection performance. The BiLSTM- based emotion-aware architecture achieves the best overall performance on MMSD2.0 with 83.12% accuracy, 81.10% precision, 85.07% recall, and 83.04% F1-score, while the GCN-based approach achieves competitive results with 81.29% accuracy and 81.97% F1- score. Cross-dataset evaluation reveals robust generalization capabilities, with the BiL- STM model maintaining 79.86% accuracy on MMSD Original and 80.45% accuracy on MEMOTION, demonstrating effective transferability of emotion-aware features across dif- ferent data distributions and even different domains. These findings establish the first em- pirical evidence for the effectiveness of dual-modality emotion integration in multimodal sarcasm detection, providing a robust foundation for future research in emotion-aware approaches to understanding digital communication nuances.

2025-05-01T00:00:00Z TYAGI, HARSHIT http://dspace.dtu.ac.in:8080/jspui/handle/repository/22521 2025-12-29T08:45:40Z 2025-06-01T00:00:00Z

Title: SECURE LIGHTWEIGHT AUTHENTICATION FOR INTERNET OF THINGS Authors: TYAGI, HARSHIT Abstract: The proliferation of Internet of Things (IoT) devices across various domains has in- troduced new challenges in ensuring secure and efficient communication over inherently insecure networks. Authentication protocols in such environments must balance robust- ness, lightweight execution, and resilience against evolving attack vectors. Given the limitations of conventional schemes in resource-constrained and high-risk settings, this thesis report presents two novel contributions designed to enhance authentication security in IoT ecosystems through cryptographic and architectural innovations. As part of this effort, the first contribution targets security enhancement in Internet- of-Medical-Things (IoMT) scenarios. Robust schemes are particularly critical in such settings due to the transmitted data’s sensitivity and resource-constrained device limi- tations. While Masud et al. proposed a protocol for securing data in IoMT networks, their approach remains vulnerable to offline password-guessing and privileged insider at- tacks, posing serious privacy and patient safety risks. To address these issues, this report proposes a novel protocol, P-MASFEP (security-enhanced PUF (Physically Unclonable Functions)-based Mutual Authentication & Session key establishment using Fuzzy Ex- tractor & PKI (Public Key Infrastructure)). P-MASFEP integrates PUFs with fuzzy extractors to actively derive stable cryptographic keys from biometric input, mitigating password-guessing risks. It also employs PKI to distribute session keys securely and ensures protection against insider threats through mutual authentication. The second contribution focuses on overcoming the inherent limitations of a traditional authentication framework, Kerberos. Its traditional design faces challenges in resource- constrained IoT environments, including computational inefficiencies, lack of clock syn- chronization, and limited scalability. In addition to these limitations, Kerberos remains vulnerable to several modern attacks such as password-guessing, Kerberoasting, Golden Ticket, and Silver Ticket attacks. Prapty et al.’s KESIC, adapts Kerberos for IoT by introducing optimizations. However, it relies on symmetric cryptography for authentica- tion and key exchange. Additionally, it remains susceptible to password-based attacks, necessitating a more secure approach. This work proposes two novel protocols to address these issues: (1) Kerberos with FIDO (Fast Identity Online) Integration (KFI), which integrates FIDO’s passwordless authentication to eliminate password-derived vulnerabil- ities; and (2) Kerberos with FIDO and Lightweight extension for IoT (KFLIT), which extends KFI by incorporating lightweight HMAC and XOR operations to reduce compu- tational overhead, counter-based synchronization to eliminate dependency on real-time clocks, and an attestation mechanism to verify IoT device integrity before granting access. Together, the proposed solutions address critical gaps in current authentication mech- anisms for constrained environments. By tackling domain-specific (IoMT) and general- purpose (IoT) challenges, this report contributes to building a secure and scalable au- thentication foundation for next-generation connected systems.

2025-06-01T00:00:00Z