Android malware is also changing fast and is becoming problematic to conventional static and dynamic detection techniques that tend to ignore real-time API communications, log patterns, and access patterns. The current paper suggests a new Android malware detection and classification system based on the combination of Graph Convolutional Neural Networks (GCN) and multi-source data fusion comprising of logcat outputs, API calls, permissions, device logs, and debug messages. The GCN outputs are used to optimize Analytical Hierarchy Process (AHP) weights and Deep Q-GAN (DQ-GAN) is used to improve malware classification. The proposed model is also more precise, accurate, recall, AUC, and specificity than detection delay, showing a better adaptability and strength to emerging Android malware threats.