Abstract
Compared with traditional compressed spectrum sensing, adaptive compressed spectrum sensing can greatly reduce the sampling rate. However, it requires complex optimization algorithms for signal reconstruction and extra samples to validate the accuracy of the reconstructed signal, which makes real-time signal processing impossible. In order to further reduce the number of samples and enable low complexity greedy algorithms, e.g., orthogonal matching pursuit (OMP), in the adaptive compressed spectrum sensing, we propose a cross validation (CV) OMP algorithm (CV-OMP) for the adaptive compressed spectrum sensing. The proposed CV-OMP algorithm embeds the CV process into an OMP algorithm, so it does not require extra samples as testing set to validate the accuracy of the reconstructed signal. Furthermore, the accuracy of the reconstructed signal is verified by the CV, which is used as the criterion to stop the iteration, so that the proposed CV-OMP algorithm does not require the prior knowledge of sparsity order for signal reconstruction. Simulation results indicate that the proposed CV-OMP algorithm achieves better reconstruction performance than that of conventional OMP algorithms, and also effectively reduce the number of samples in the adaptive compressed spectrum sensing.