Wireless Sensor Networks (WSNs) have a wide variety military applications including battlefield surveillance, enemy tracking, and target classification. In this paper, we propose a WSN architecture based on a mobile sink. The proposed architecture differentiates the regular data from the critical data, and leverages this difference to reduce the energy consumption in WSNs. We formally modeled the mobile and static sink based architectures and analyzed the energy consumption and data delays using a probabilistic model checker.