Résumé
We investigate the problem of resilient and energy-aware Virtual Network Function (VNF) placement and routing in softwarized networks under the threat of targeted cyberattacks. We model the system as a bilevel interdiction game, where a malicious attacker strategically disrupts servers within a fixed resource budget, while a network provider reacts by minimizing energy consumption through optimized VNF deployment and flow routing. The lower-level problem includes capacity constraints, service function chaining, and a server energy model accounting for idle and load-dependent consumption. Attack-induced load shifts are captured via additive energy penalties on compromised nodes. To solve this inherently difficult bilevel integer program, we de-velop a single-level reformulation via interdiction cuts and propose a cutting-plane algorithm to explore the attacker's strategy space efficiently. Numerical experiments show the effectiveness of the approach in quantifying trade-offs between resilience and energy efficiency, supporting trustworthy and adaptive NFV deployment in critical infrastructures.