DigiCaM: Digital Twin–Assisted Cognitive Spectrum Sharing for 6G Coexistence with Passive FSS-ES Receivers in 12 GHz Band

The lower 12 GHz band (12.2-12.7 GHz) has emerged as a key candidate for future 6G deployments, yet its use is constrained by the need to protect highly sensitive passive Fixed Satellite Service receive-only Earth Stations (FSS-ES), which neither transmit nor provide interference feedback. This absence of incumbent sensing or reporting fundamentally limits conventional dynamic spectrum sharing approaches and creates a challenging cognitive networking problem. In order to address this problem, this paper presents DigiCaM, a Digital Twin-assisted cognitive spectrum sharing framework that enables feedback-free coexistence between 6G macro base stations (MBSs) and passive FSS-ES receivers in the 12 GHz band. DigiCaM integrates a context-aware Digital Twin (CDT) that models site-specific propagation, 3D building geometry, and weather dynamics to infer interference without incumbent feedback, together with intelligent MBS control algorithms. We formulate the coexistence problem as a constrained optimization and show it is NP-hard. To address this, we develop (i) a greedy algorithm with provable (1 − 1 e) approximation guarantees via monotone submodular maximization, and (ii) a recurrent deep reinforcement learning (R-DRL) approach with Lagrangian duality to enforce regulatory interference constraints while adapting to dynamic network and environmental conditions. Extensive evaluations in a realistic 33-MBS deployment in Blacksburg, VA demonstrate that DigiCaM achieves interference estimation accuracy comparable to highfidelity ray tracing while reducing computation time by orders of magnitude. Compared to the state-of-the-art baselines, DigiCaM (R-DRL) and DigiCaM (Greedy) enable up to 63.16% and 21.05% more active MBSs, respectively, and deliver spectral efficiency improvements of up to 84% and 26.67%, while strictly meeting FSS-ES protection constraints.