Abstract: Distributed space-based early warning radar systems enable large-scale， high-precision space target monitoring through multi-platform collaboration， addressing the critical need for comprehensive situational awareness in modern warfare. However， while the distributed architecture enhances system performance， it introduces the technical challenge of grating lobe clutter suppression. This study systematically investigates clutter suppression methods based on the frequency division multiple-input multiple-output （FD-MIMO） technique to address this issue. First， a complete space-time echo signal model is established by analyzing the spatial geometry and operational characteristics of distributed space-based radar. The intrinsic physical mechanism by which FD-MIMO suppresses grating lobes via carrier frequency diversity is examined in detail through array pattern function analysis. Second， considering the complex electromagnetic environment in practical applications， three typical scenarios are defined based on echo correlation characteristics： （1） both clutter and target are correlated， （2） clutter is decorrelated while the target remains correlated， and （3） both clutter and target are decorrelated. Corresponding to these scenarios， three grating lobe suppression approaches are rigorously investigated： （1） space-time-frequency adaptive processing， （2） single-carrier space-time adaptive processing （STAP） cascaded with multi-carrier coherent processing， and （3） single-carrier STAP cascaded with multi-carrier non-coherent processing. The phase relationships between echoes at different carrier frequencies are mathematically derived to support these methods. Comprehensive simulations are conducted to validate the effectiveness of FD-MIMO-based grating lobe suppression. The performance of the three methods is comparatively analyzed in terms of clutter rejection capability， computational complexity， and robustness. The results demonstrate that space-time-frequency adaptive processing achieves the best overall performance and highest robustness， albeit with high computational cost and strict echo correlation requirements. In contrast， single-carrier STAP with multi-carrier coherent processing offers near-optimal performance while maintaining lower computational complexity and looser requirements on clutter correlation， making it more suitable for practical engineering applications. This study provides both fundamental theoretical support and practical engineering guidance for designing distributed space-based early warning radar systems. The findings confirm that FD-MIMO techniques effectively mitigate grating lobe effects while offering flexible implementation options that balance performance and complexity. The proposed framework is a valuable reference for future developments in advanced space surveillance architectures.