Journal of Networking Technology Tuan Nguyen Minh 17 2 2026 https://doi.org/10.6025/jnt/2026/17/2/47-62 https://www.dline.info/jnt/fulltext/v17n2/jntv17n2_1.pdf The rapid expansion of large-scale complex networks has intensified the need to understand how structural resilience and functional diffusion processes interact under perturbations. This paper presents a unified framework that integrates percolation-theoretic robustness metrics with canonical influence propagation models, addressing a critical gap in network science where structural stability and information flow are typically analyzed in isolation. We formalize network robustness through the persistence of the giant connected component, global efficiency, and a scalar robustness index, evaluating degradation under both random failures and targeted attacks. Analytical foundations, including the Molloy Reed criterion and kshell decomposition, are coupled with simulation-based protocols to quantify critical percolation thresholds and fragmentation dynamics. Concurrently, we model diffusion using the Independent Cascade and Linear Threshold frameworks, revealing inherent trade-offs between structural resilience and propagation efficiency. Hubs accelerate the spread of information but represent critical vulnerabilities, while high clustering enhances local fault tolerance yet may impede global diffusion. Through empirical analysis of realworld scientific collaboration networks, we demonstrate that optimal network design requires contextaware balancing of redundancy, centrality distribution, and connectivity patterns. The study establishes practical design principles for engineering adaptive architectures that withstand stochastic disruptions and adversarial attacks while maintaining functional diffusivity. Ultimately, this framework provides actionable insights for developing resilient infrastructure, robust social platforms, and efficient communication systems, with proposed future extensions to temporal and multiplex network dynamics.