Analysis and Identification of Malicious Applications

The "Malware App Detection System" project enhances mobile security by detecting malicious apps through machine learning (ML) analysis of app behaviours. It examines behaviours like system calls (requests apps make to the operating system), permissions (levels of access apps have), and network connections (internet interactions). This focus on behaviour, rather than relying on specific malware signatures, enables the system to identify both known and new threats. The detection system is built using a dataset of both benign (harmless) and malicious apps, with each app’s behaviour categorized and labelled. By analysing system calls, the system identifies unusual requests, which may indicate malicious intent. Permissions are also reviewed, as apps that seek abnormal access levels, like to contacts or storage, might have harmful intentions. Network connections are monitored to detect suspicious activities, such as connections to unauthorized servers, which could indicate data theft. To achieve this, ML models like Random Forest are trained on these behaviour patterns, learning to distinguish between safe and harmful apps. Once deployed, the system provides real-time detection, alerting users if it finds potentially dangerous activities. This proactive, behaviour-based approach protects users from unauthorized access and data leaks, adapting to new malware types as they arise. In doing so, the project offers a flexible and effective solution to the evolving challenges of mobile security.