<p>The intricate nature of modern edge architectures, relying on a vast array of computational logic and lightweight communication protocols, creates vulnerabilities that expose them to a broad spectrum of security threats. Moreover, security vulnerabilities can significantly impact the energy footprint of edge servers in these architectures. Our approach utilizes the concurrent stochastic game (CSG) formalism to model the behavior of IoT communication entities (players) while accounting for potential attacks at the communication edge and the resulting energy consumption caused by such attacks. We rely on the PRISM-games language for automated analysis where the game goals modeling functional and security requirements are expressed using reward probabilistic alternating temporal logic (rPATL). To validate our approach, we examine a data corruption attack applied to dam water flow control and study its side effect on energy consumption associated with SensiNact gateways. Our key innovation lies in using formal models at the architectural level to explore potential attacks. These models capture synchronous and asynchronous communication styles, along with their associated energy consumption. The methodology and the implemented formalism offer a significant advancement over traditional game equation models while still achieving the desired security and energy evaluation. Numerical results show that compared to synchronous communication, asynchronous styles suffer from significantly larger infected buffers and higher energy consumption due to attacks ranging from 66 to 91%.</p>