Mohamed El-Dosuky

Dr. Mohamed El-Dosuky


Research Focus & Metrics

Academic Career Length

3Years (2024-Present)

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

Representative Publications

View All Publications
23 March 2025 in The Journal of Supercomputing
Verified
Journal Article
Open access
PDF file
EEEman Ibrahim Abd El-Latif
Eman Ibrahim Abd El-Latif
MEMohamed El-Dosuky
Mohamed El-Dosuky
Developing an effective energy consumption prediction model has become a central focus in drone-related research. As a result, numerous models are proposed, each differing in complexity and focusing on factors such as thrust and environmental conditions. This paper presents two mathematical models and a machine learning approach to estimate the energy consumption of drones. The Eldosuky model, the first one, examines the effects of weight, thrust, and air-rotor interaction on power consumption. The second model, Eman, takes into account the drone’s mass, payload, and external factors like altitude and airspeed. These models are used to simulate drone power consumption in a variety of scenarios, demonstrating their accuracy and utility in real-world situations. Additionally, this paper uses a random forest regressor, a machine learning model that uses actual data to validate energy predictions and simulate drone performance. Then, the accuracy of the model is improved by applying the Fick’s law algorithm optimizer, which contains three phases of motion. These phases are referred to as diffusion operator (DO), equilibrium operator (EO), and steady-state operator (SSO). Evaluation metrics are compared both before and after normalization for seven models: Eldosuky, Eman, D’Andrea, Dorling, Stolaroff, Kirchstein, and Tseng. The Eldosuky model has a lower RMSE (500.7558) and MAE (313.3711) before normalization. The Fick’s law algorithm optimizer exhibits optimal metrics following normalization, showcasing a noteworthy enhancement with RMSE of 0.24303 and MAE of 0.08805.

1 May 2025 in Journal of Computer Science
Verified
Journal Article
Open access
PDF file
ANAlaa Najmi
Alaa Najmi
MEMohamed El-Dosuky
Mohamed El-Dosuky
Effective testing scenarios are necessary to guarantee the dependability and Caliber of software. Conventional techniques for creating these scenarios frequently involve a great deal of manual labor and might not fully cover all software requirements. In order to improve the automation and Caliber of software testing scenario development, this study investigates the combination of Natural Language Processing (NLP) and Deep Learning (DL) approaches with ChatGPT, an advanced language model by OpenAI. The suggested method automatically creates a variety of thorough test cases by utilizing ChatGPT's sophisticated natural language processing capabilities. To evaluate the model's capacity to comprehend intricate software requirements and generate pertinent situations, a comparison between conventional scenario-generation techniques and those improved by ChatGPT is carried out. The process is divided into four stages: Requirement parsing, in which natural language software requirements are analyzed and interpreted using NLP models; scenario generation, in which a transformer-based model is used to generate testing scenarios that are logical and appropriate for the environment. an automation pipeline that uses Hugging Face Transformers and Python to speed up the scenario generating process and evaluation metrics that evaluate the created scenarios according to requirement coverage and relevance coherence. The effectiveness of this method is illustrated through a case study on evaluating an Optical Character Recognition (OCR) system for private documents. The results show that integrating ChatGPT with NLP and DL greatly enhances the depth of testing scenarios, speeds up the generation process, and lowers manual labor. The potential of ChatGPT to automate and optimize software testing is demonstrated in this study, providing a more effective and flexible solution for a variety of testing scenarios.

27 June 2024 in Journal of Ambient Intelligence and Humanized Computing
Verified
Journal Article
Closed access
PDF file
EEEman Ibrahim Abd El-Latif
Eman Ibrahim Abd El-Latif
MEMohamed El-Dosuky
Mohamed El-Dosuky
ADAshraf Darwish
Ashraf Darwish
Abou ELLa HassanienAbou ELLa Hassanien
Abou ELLa Hassanien

1 November 2025 in Journal of Computer Science
Verified
Journal Article
Open access
PDF file
NGNoman Aasif Gudur
Noman Aasif Gudur
MEMohamed El-Dosuky
Mohamed El-Dosuky
SKSherif Kamel
Sherif Kamel
The increasing volume of consumer data necessitates reliable edge devices for personalized user experiences. Federated Learning (FL) offers a state-of-the-art approach to decentralized machine learning by leveraging data distributed across multiple client devices. However, user data privacy remains vulnerable to corruption through feature heterogeneity and malicious attacks. While several privacy-preserving techniques have been previously implemented, they suffer from implementation constraints and limited robustness against sophisticated attacks. This paper proposes a deep convolutional neural network mechanism that enhances privacy preservation in FL by combining Homomorphic Encryption (HME) and Secure Multiparty Computation (SMC). The proposed approach is validated through model verification on the CIFAR-100 dataset and a healthcare diabetes dataset case study. Results demonstrate that the proposed mechanism outperforms existing privacy protection methods, particularly against backdoor attacks. By ensuring stronger privacy guarantees, this approach facilitates broader adoption of FL technology across privacy-sensitive domains.

6 October 2025 in arXiv
Verified
Preprint
Open access
PDF file
MEMohamed El-Dosuky
Mohamed El-Dosuky
The Internet of Things (IoT) has revolutionized various applications including agriculture, but it still faces challenges in data collection and understanding. This paper proposes a real-time framework with three additional semantic layers to help IoT devices and sensors comprehend data meaning and source. The framework consists of six layers: perception, semantic annotation, interoperability, transportation, semantic reasoning, and application, suitable for dynamic environments. Sensors collect data in the form of voltage, which is then processed by microprocessors or microcontrollers in the semantic annotation and preprocessing layer. Metadata is added to the raw data, including the purpose, ID number, and application. Two semantic algorithms are proposed in the semantic interoperability and ontologies layer: the interoperability semantic algorithm for standardizing file types and the synonym identification algorithm for identifying synonyms. In the transportation layer, raw data and metadata are sent to other IoT devices or cloud computing platforms using techniques like WiFi, Zigbee networks, Bluetooth, and mobile communication networks. A semantic reasoning layer is proposed to infer new knowledge from the existing data, using fuzzy logic, Dempster-Shafer theory, and Bayesian networks. A Graphical User Interface (GUI) is proposed in the application layer to help users communicate with and monitor IoT sensors, devices, and new knowledge inferred. This framework provides a robust solution for managing IoT data, ensuring semantic completeness, and enabling real-time knowledge inference. The integration of uncertainty reasoning methods and semantic interoperability techniques makes this framework a valuable tool for advancing IoT applications in general and in agriculture in particular.