A User Threat-Driven Multi-Objective Adaptive VM Allocation Framework for Cloud Security

Cloud computing systems face side-channel attacks (SCAs, where attackers exploit information leaked through physical resources like CPU cache to compromise VMs on shared servers) when multiple VMs run on the same physical machine. This paper proposes a framework that analyzes user behavior patterns to intelligently place VMs across servers and then continuously moves them around to prevent attackers from keeping their malicious VM next to a target VM long enough to launch an attack, achieving up to 25% reduction in attack risk while improving resource efficiency.