<p>Fog computing is a dynamic, distributed infrastructure that extends cloud services-such as storage, computing, and communication-to the edge of the network. However, since fog nodes are located close to end users and may handle sensitive data, they must be trustworthy. Traditional cryptographic methods are insufficient to defend against internal attacks, such as those posed by rogue nodes that have already been authenticated and integrated into the system. Existing approaches have largely overlooked collusion attacks, which can compromise the accuracy of trust evaluations. Only a few methods offer limited resistance to such attacks, while others address subsets of collusion that are significantly weaker than collusion itself. Moreover, environments like fog computing are inherently dynamic, with constantly changing network topologies. This variability must be considered in any trust model. To tackle these challenges, we propose a collusion-resistant trust management system for secure fog-to-fog collaboration. Our system leverages quality of service (QoS) and quality of security (QoSec) metrics for direct trust and dynamically incorporates neighbor recommendations to evaluate indirect trust. In our recommendation model, we introduce a mechanism to incentivize neighbors to provide accurate information by evaluating quality of recommendation (QoRec), thereby reducing the risk of collusion. Finally, our trust aggregation is based on a weighted sum that dynamically balances direct and indirect trust. Our results demonstrate that the proposed approach can withstand the presence of up to 80% malicious nodes, and we have achieved a 44% improvement in convergence speed for trust computation compared to the previous related work.</p>