Seif ElDein Mohamed, Mostafa Ashraf, Amr Ehab, Omar Shereef, Dr.Eslam Amer, Eng.Haytham Metawie, Eng.Mostafa Badr
December 28, 2020
Nowadays, the mobile industry is in rapid evolution making smartphones available with affordable rates for all segments of society. Smartphones’ purposes are not limited to making phone calls or sending messaging, users can also take photos, store personal data, do online banking and trace their daily activities. The more applications appear, the more security becomes a concern to mobile users. This concern arises from the fear of being subjected to a security breach that jeopardizes confidential personal data such as emails, passwords, location, credentials etc. Malware applications which are developed for the sake of compromising users’ personal data are also increasing rapidly day after day. In our work, we aim to design an intelligent detection framework for Android malware applications. The framework uses different analysis-based approaches along with different machine learning algorithms to distinguish between benign and malicious
1.1 Purpose of this document
The main purpose of this Software Requirements Specification document is to outline the fundamentalrequirements of our system ’Detecting Malicious Android Applications based on permissions using Machine Learning Algorithms’. The system aims to detect any malware in Android device systems to protect our clients’ system. Our model will be implemented in a mobile application that runs on Android systems. We also provide a fulfilled description of each single stage input, Post-Condition, and algorithms used in this stage. Along with a full illustration for each stage’s requirements and development process.
1.2 Scope of this document
This document targets the clients’ and companies’ administration which has a role in the companies’ business flow. The document provides the detailed functional and non-functional requirements, as well as the main functionalities of our system presented in the different stages of the system development.
1.3 System Overview
The development of the system is split into 4 stages as follows:
1-The input stage, in which the data is priority, collected From DefenseDroid that contains unique Android applications. Also, it has permissions for benign and malicious applications.
2-In pre-processing phase, Normalization and standardization will also be applied on the data-set then extract a binary image.
3-In processing stage, the data-set is split into two subsets training and testing respectively. We focus on training the different models using training set, our malware detection model will be trained and tested on the data-set by applying CNN methods.
4-In the last stage, we will classify each selected application as either benign, malware.
1.4 System Scope
The proposed system classifies malware behaviours of Android Applications with efficiency and viability, utilizing machine learning techniques and dynamic behaviour analysis to naturally identify malware in all file types.