Driven by 6G communication technologies and the explosive growth of low Earth orbit (LEO) satellites, non-cooperative satellites—such as the expanding Starlink and Starshield constellations, which dominate global communication networks—pose emerging spatial security threats to national interests. Non-cooperative satellites are defined as those that do not actively communicate with ground-based control systems, typically lacking signal-emitting devices for tracking or identification, or intentionally concealing operational information. Ground-based mission planning involves the integrated scheduling of non-cooperative satellite observations under constraints such as satellite quantity, mission requirements, and resource limitations. This process establishes a mapping between ground resources and observation tasks to achieve optimal resource allocation. The resulting challenges in non-cooperative satellite ground-based observation task planning have become a critical bottleneck in advancing space situational awareness capabilities. Existing research, particularly on large-scale non-cooperative satellites with active evasion capabilities, faces significant technical barriers in computational efficiency, dynamic task responsiveness, and multi-objective optimization. This study systematically examines the architectural features of planning models and the technological evolution of optimization algorithms across three satellite cluster scales (small, medium, and large). By analyzing these dimensions, we identify key challenges and future directions in the field, offering theoretical foundations and methodological frameworks for advancing ground-based mission planning methodologies amid intensifying orbital resource competition.