Automotive User Interface Based on LSTM-Grid Search Deep Learning Model for IoT Security Change Request Classification

The field of the Internet of Things (IoT) is rapidly growing in significance, with roughly fifty billion devices being used in technology for computing by the end of 2020. However, the interdependence of IoT devices, as well as the variety of components used in their implementation, has caused a variety of issues, such as insufficient testing of change requests (CR) that affect security requirements. One way to address these security issues is to provide a deeper classification of security requirements that have an impact on the overall software process, such as software testing. Thus, the primary goal of this study is to assist software testers in prioritizing changes requested to enhance software security on IoT-based devices. Therefore, a deep learning-based approach to security CR classification is proposed. In this study, the Long Short Term Memory (LSTM) model is used and enhanced through the grid search tuning method. The LSTM-based grid search deep learning classifier identifies the class (i.e., sub-characteristics defined by the ISO 25010 quality model) of a given security CR from IoT-based devices with an average classification accuracy of 79%. Finally, to validate the robustness of our proposed research methodology, we developed an automated classification user interface for software testers.