The integration of the Over-the-Air Computation (AirComp) in Concentrated Solar Power (CSP) systems offers a transformative potential for real-time data aggregation in 6G-IoT networks. However, such an innovative deployment introduces critical vulnerabilities to jamming attacks that distort the aggregated signals without detection. This paper introduces a robust security framework that leverages an artificial intelligence (AI) statistical signal analysis, adaptive detection thresholds, and spatially-aware mitigation to detect and eject both simple and coordinated jamming attacks in real-world AirComp CSP systems. Detailed experimental validation at the National Solar Thermal Test Facility demonstrates significant improvements in the data aggregation and attack detection rates. It also ensures trustworthy data aggregation from commercial CSP fields, while maintaining their operational resilience.

Integrated Sensing and Communication (ISAC) has become critical in public safety and disaster response operations given its ability to jointly support real-time data collection and efficient communication. In this paper, the DRAGON framework is introduced to address existing research gaps by introducing a reinforcement learning-based approach that enables the users to autonomously select UAVs for communication and optimize the resource allocation and task incentives allocation. A Stackelberg game-theoretic approach enables the DRAGON framework to ensure the users' efficient sensing and communication in disaster areas by prioritizing both the users' utility and the UAVs' operational energy efficiency. Comprehensive experiments show the superior performance of DRAGON over other existing models and also validate its scalability and real-world applicability.