The College of Education for Pure Sciences, Department of Computer Science, presented a master's thesis on deep learning to improve the prevention and detection of black hole attacks in wireless multimedia sensor networks.
The thesis, presented by researcher Zakaa Ali Kazim, aims to enhance the security of wireless multimedia sensor networks (WMSNs) by introducing an integrated hybrid framework. This framework combines deep learning-based anomaly detection techniques with a reputation-based data retrieval and routing (R-DRAFT) mechanism to improve network efficiency and security. This framework is important for multimedia applications in dynamic environments where network security is critical.

The thesis concluded that the proposed system demonstrated superior performance in terms of packet delivery rate, reduced data loss, end-to-end delay, and increased transmission rate. These improvements were achieved by using the SA-DCBiGNet model, based on a dual-attention mechanism for accurate detection of malicious nodes, and trust management via a dynamic reputation table. This framework provides a scalable and secure solution for multimedia data transmission in WMSNs, making it suitable for multimedia applications in dynamic environments

.