Deep Learning Aproach for Fault Prediction in IoT Systems

Please enable javascript in your browser.

Hosseini Nodeh, Ali | 2025

199 Viewed

Type of Document: M.Sc. Thesis
Language: Farsi
Document No: 58552 (19)
University: Sharif University of Technology
Department: Computer Engineering
Advisor(s): Fazli, Mohammad Amin; Habibi, Jafar
Abstract:
The growth of heterogeneous Internet of Things (IoT) systems with limited resources has made predicting software bugs a critical challenge for maintaining security and stability, as bugs and errors can lead to widespread failures in the real world. Although many bugs stem from programming errors and dependencies, the complexity of IoT demands innovative intelligent solutions. In this context, deep learning approaches have shown great potential for identifying errors in these heterogeneous systems due to their ability to understand complex semantic and syntactic patterns in code. This thesis systematically examines these novel methods; it first describes the specific challenges and bugs related to the IoT, then provides a detailed analysis of deep learning techniques, and finally offers a comprehensive overview of the current state and future of this research area, while also discussing methods for collecting specialized data to train these models
Keywords:
Internet of Things ; Defect Prediction ; Deep Learning ; Software Engineering ; Fault Detection ; Failure Analysis