The attack mitigation technique, presented as graph-partition scheme, enhances the resistance of dense mesh communication network to the fast spreading cyber-attacks. The research work presents an adaptive graph-partitioning framework that incorporates Louvain’s community detection, METIS’s multi-level partitioning and multi-objective optimization for optimal attack containment. The framework is designed to achieve minimal reverse propagation while maximizing latency, connectivity, load balancing and routing stability in dynamic environments. The Pareto configurations of partitions are obtained by using NSGA-II and Particle Swarm Optimization. Dynamic adaptive partitioning also allows for the protection against wormhole, Sybil and DDoS attacks. Experimental results show that a real-time distributed mesh infrastructure becomes more scalable, resilient and secure with low communication overheads.