Ubaid Ullah

Ubaid Ullah

Auto-generated profile, based on publicly available information

Ubaid Ullah

Department of Computer Science and Information Technology, BRAINS Institute, Peshawar, Pakistan

Research Focus & Metrics

Academic Career Length

5Years (2022-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
17 February 2022 in IEEE Access
Verified
Journal Article
Open access
ISIsmail
Ismail
Muhammad Ismail MohmandMuhammad Ismail Mohmand
Muhammad Ismail Mohmand
HHHameed Hussain
Hameed Hussain
AKAyaz Ali Khan
Ayaz Ali Khan
UUUbaid Ullah
Ubaid Ullah
MHMuhammad Haleem
Muhammad Haleem
Distributed network attacks are referred to, usually, as Distributed Denial of Service (DDoS) attacks. These attacks take advantage of specific limitations that apply to any arrangement asset, such as the framework of the authorized organization’s site. In the existing research study, the author worked on an old KDD dataset. It is necessary to work with the latest dataset to identify the current state of DDoS attacks. This paper, used a machine learning approach for DDoS attack types classification and prediction. For this purpose, used Random Forest and XGBoost classification algorithms. To access the research proposed a complete framework for DDoS attacks prediction. For the proposed work, the UNWS-np-15 dataset was extracted from the GitHub repository and Python was used as a simulator. After applying the machine learning models, we generated a confusion matrix for identification of the model performance. In the first classification, the results showed that both Precision (PR) and Recall (RE) are $\sim 89$ % for the Random Forest algorithm. The average Accuracy (AC) of our proposed model is ~89% which is superb and enough good. In the second classification, the results showed that both Precision (PR) and Recall (RE) are approximately 90% for the XGBoost algorithm. The average Accuracy (AC) of our suggested model is ~90%. By comparing our work to the existing research works, the accuracy of the defect determination was significantly improved which is approximately 85% and 79%, respectively.

16 December 2022 in Security and Communication Networks
Verified
Journal Article
Open access
PDF file
UUUbaid Ullah
Ubaid Ullah
UMUsama Musharaf
Usama Musharaf
MHMuhammad Haleem
Muhammad Haleem
The modeling of software functional requirements is very important in the development of software systems. In addition, it is also highly important to model security requirements, specifically authentication requirements, to prevent security risks and threats at the design level. The authentication requirements impose various constraints on granting access to only legitimate users to use computer resources. These constraints play an important role in the modeling of functional and authentication concerns of software systems. However, due to their crosscutting nature, their modeling results in pervasiveness across several design models. These constraints crosscut other constraints, which raises the problems of tangling and scattering. As a result, it is highly difficult to manage and maintain the constraints put on design models. Moreover, taking ad hoc approaches to deal with these constraints in complex systems is likely to result in faulty design models. All the existing approaches only deal with crosscutting behavior and completely ignore the crosscutting constraints. Therefore, our main research objective is to improve the modeling of crosscutting constraints and remove the scattering and tangling problems. For this purpose, we propose an authentication-oriented approach to modeling these constraints using the aspect-oriented technique. Using this approach, two case studies are implemented. Finally, the results show that the tangling and scattering problems are eliminated by separately modeling the crosscutting constraints as aspects using aspect OCL. This reduces the overall number of constraints and lowers the maintenance and management effort, which ultimately reduces the complexity of sequence models. The constraint-oriented sequence models are successfully verified, which shows that the output models are correct and complete. To conclude, our research approach is more useful and effective than the other approaches as it covers all the necessary steps required for functional and authentication behavior and constraint modeling.

1 January 2026 in International Journal of Intelligent Systems
Verified
Journal Article
Open access
UUUbaid Ullah
Ubaid Ullah
SLSonia Laudanna
Sonia Laudanna
ASAndrea Di Sorbo
Andrea Di Sorbo
CVCorrado Aaron Visaggio
Corrado Aaron Visaggio
The emergence of large language models (LLMs) has significantly advanced natural language processing (NLP); however, their capacity to generate human‐like content introduces serious security concerns. In particular, the misuse of LLMs for disinformation and impersonation on social media platforms such as X creates new opportunities for large‐scale manipulation and deception of users. This study aims to conduct a comprehensive investigation to (i) understand the distinct stylistic features effectively mimicked by 10 different LLMs and (ii) distinguish between LLM‐driven and human authors when LLMs are explicitly instructed to mimic a specific human writing style. In particular, we design adversarial prompts to mimic the writing style of human authors based on key stylometric features (quantitative analysis of writing style) and assess the mimicking effectiveness of different LLMs through extensive statistical testing. In addition, we conduct a survey that gauges human ability to recognize the author of a text and train machine learning models to identify human‐ and LLM‐driven authors, focusing on scenarios where specifically crafted adversarial prompts are employed to facilitate style impersonation. Our findings demonstrate that, when explicitly instructed, LLMs can effectively replicate features of human writing style. In addition, the survey results indicate that it is challenging for the participants to distinguish between the different types of authors. In fact, the participants only demonstrated a classification accuracy of 15% in correctly identifying the text generated by the LLM. In contrast, high detection performance is achieved only when the training data incorporate adversarially generated LLM samples produced using impersonation‐oriented prompts. Under this threat‐model–aligned training regime, stylometric‐based classifiers exhibit strong discriminative capability, attaining classification accuracy of up to 99% in distinguishing human‐authored text from LLM‐generated authorship.