Machine Learning-Driven Content Popularity Prediction and Cache Optimization in D2D Clustered Networks

Collection of recent networking and edge-AI research with multiple security- and safety-relevant findings: ML-driven D2D caching and cluster-level popularity prediction (user heterogeneity and willingness to participate) that could be abused for content poisoning or privacy leakage; edge AI/IoT architectures for scalable sensing showing large reduction in upstream traffic but increasing reliance on local preprocessing; privacy-aware MARL for 6G VR slice management that reduces privacy leakage but highlights risks from agent collaboration and mobility-driven data exposure; experimental evidence