Wireless sensor nodes are, typically, resource limited. Therefore, the major functions of the Wireless Sensor Network (WSN) cannot be accomplished without collaboration among sensor nodes. In this paper, we present the Wireless Sensor Networks Supported Cooperative Work (WSNSCW) model. The key contribution of this model, when comparing to other models, is allowing for the representation of all the components and properties of a WSN. We also present the main entities and properties of this graph-based model, as well as its formalization.

This study focuses on network behavior analysis (NBA) by capturing and transforming user traffic. DDoS attacks and malicious network behavior are known to cause severe security and efficiency issues. Previous research has demonstrated the effectiveness of various anomaly detection techniques in networks, but this approach extends these studies by incorporating a greater number of variables for a more detailed analysis of user behavior, relating it to principles from behavioral sciences. The objective of this research is to develop a system that, through data transformation and machine learning, can detect profiles of potential attack sources, assisting network administrators in streamlining audit processes and equipment configuration reviews. Additionally, this system could serve as a forensic computing solution by detecting a historical pattern of the individual’s behavior, thereby providing deeper insights into potential security threats and aiding in the investigation of past incidents. To achieve this, data from the public BCCC-Dataset, BCCC CPacket Cloud-based DDoS 2024, Behaviour-Centric Cybersecurity Center (BCCC), containing benign and malicious activities, were used. These data were transformed using one-hot encoding and grayscale images representing IP behavior were generated. Using NBA techniques, deep learning algorithms with ResNet architectures were applied to analyze these images and predict potential threats. Grad-CAM importance maps were utilized to highlight the variables that define risk in a profile. The results show that the model can identify suspicious behavior patterns with high accuracy, significantly reducing response time to security incidents. The implications of these findings are multiple: network management efficiency is enhanced through automated threat detection and reduced false positives. Decision-making is optimized thanks to the system’s predictive intelligence, providing precise and detailed data-driven insights. Operational costs are reduced by minimizing losses from incidents and improving resource efficiency. This approach enhances network security and offers a powerful tool for proactive threat management, informed by behavioral sciences. <!----> <!----> <!----> <!---->