ProFlow: RL-Driven and Performance-Aware Proactive Flow Placement in Datacenter Networks

The document is an arXiv networking research feed covering proactive datacenter congestion management, workload-inference side channels through shared RTT behavior, Mixture-of-Experts incast mitigation, empirical networking tooling, auditable AI for 6G, intelligent resource allocation, reconfigurable wireless systems, semantic communications, agentic-network reliability, and adaptive AI-based TCP. The primary security-relevant finding is that shared congestion and latency observations can leak information about co-located tenant workloads despite logical network isolation, with reported run-in