Space Domain Awareness (SDA) is fundamental to ensuring the security and sustainability of operations beyond Earth orbit, particularly within cislunar and deep-space environments. As international and commercial interest in lunar and deep-space missions accelerates, maintaining situational awareness of orbital activities, space traffic, and potential hazards has become a strategic imperative. This study introduces a resilient SDA architecture that integrates conventional tracking and identification functions with adaptive recovery mechanisms capable of operating under environmental perturbations and adversarial conditions. The proposed framework employs a graph-based modeling paradigm, wherein satellites, sensors, and targets are represented as dynamically interacting nodes, allowing the network to simulate system behavior under varying operational scenarios. The design process employs a multi-objective evolutionary algorithm inspired by biological adaptation to optimize performance, cost, and resilience concurrently. Simulation outcomes demonstrate that the architecture preserves high coverage and operational continuity even under partial degradation, validating the resilience-oriented approach. While large-scale optimization introduces increased computational demands, the proposed model achieves notable improvements in adaptability, efficiency, and robustness. Overall, the findings underscore the necessity of resilience-driven design principles for future SDA architectures tasked with supporting cislunar surveillance and protecting deep-space missions.