Survey on adaptive AI techniques for QoS and privacy preservation in multi-user 6G networks

The introduction of the sixth-generation (6G) wireless system will result in impressive advances in the field of communication, which will guarantee a low latency of communications, high bandwidth, and massive connectivity, especially to IoT systems. However, it is quite a challenge to maintain quality of service (QoS) homogeneity in an ultra-dense, heterogeneous and security sensitive 6G environment. Although rather fundamental, orthogonal frequency division multiple access (OFDMA) has limitations in the segmentation of the dynamic spectrum, latency, and interference avoidance in such a complex environment. This survey looks at how artificial intelligence (AI) can aid in addressing these issues, particularly with deep reinforcement learning (DRL), which can help realize such issues as adaptive optimization of transmission parameters and intelligent beamforming as well as real-time traffic scheduling. As well, the survey addresses AI-assisted security and privacy-protecting systems which are critical to the challenges of security posed by untrusted devices whilst maintaining QoS integrity. We analyze recent developments in integrated resource management, beamforming control, anomaly detection, and secure spectrum allocation within multi-user 6G IoT frameworks. By systematically analyzing existing methods, their advantages, and their limitations, this work highlights critical research gaps—including the lack of unified, real-time adaptive frameworks for joint QoS and security optimization. The survey concludes by outlining future directions aimed at developing scalable, intelligent, and robust AI-driven architectures to meet the demanding requirements of next-generation IoT-driven 6G communications.