Digital Twin Integrated Cooperative Spectrum Sensing

Abstract

Cooperative spectrum sensing (CSS) is widely used in cognitive radio networks to mitigate hidden-node problems and improve detection reliability through spatial diversity. However, most existing CSS schemes follow a classical memoryless design which is unreliable and inefficient when the spectrum environment changes faster than the calibration cycle. To address these limitations, this paper proposes a Digital Twin-driven CSS framework that maintains a continuously updated replica of the spectrum environment and sensing network. The framework couples instantaneous measurements with a persistent belief state: nodes generate calibrated soft evidence from sensed waveforms, and the digital twin assimilates this evidence through probabilistic updates to form posterior occupancy beliefs and node trust scores. The twin then predicts near-future occupancy and derives adaptive thresholds and fusion weights based on application-specific false-alarm and missed-detection costs, closing the loop by outputting an uncertainty-aware sensing policy for the next epoch. Simulation results show that the proposed Digital Twin-driven CSS improves robustness to non-stationary noise and heterogeneous node quality, sustains strong detection under mobility and fading, and reduces sensing/reporting cost through adaptive scheduling, achieving capabilities that memoryless CSS cannot provide simultaneously.