With the rapid evolution of 5G networks and the anticipated deployment of 6G systems, wireless communications increas ingly rely on advanced transmission techniques incorporating channel state information(CSI) feedback to improve spectral efficiency and transmission reliability. However, these advanced transmission mechanisms also introduce new vulnerabilities, particularly when attackers exploit interception/detection techniques to infer CSI-related signal characteristics and launch intelligent attacks. These challenges motivate the development of evolved security concepts and wireless system designs that jointly consider performance optimization and security protection at the physical layer. To address these challenges, this dissertation develops a comprehensive research framework spanning three key aspects: wireless threat modeling, security metric formulation, and secure wireless system design. First, adversarial attacks, which can be viewed as intelligently designed jamming attacks, are investigated to reveal potential vulnerabilities in CSI feedback–based wireless systems. Second, new security evaluation concepts are introduced to quantify the feasibility and effectiveness of such smart attacks in CSI-based wireless systems. Building upon the preceding security analysis, the dissertation further develops secure transmission strategies to mitigate the identified vulnerabilities and proposes an adaptive wireless system architecture that simultaneously achieves high performance and enhanced security. Overall, this work establishes a unified framework for analyzing wireless security and guiding the design of secure, high-performance communication systems for next-generation wireless networks.