The spectrum sensing algorithm for cognitive network based on LLE and random forest

Focused on the spectrum sensing in low signal-to-noise ratio, we propose a novel spectrum sensing method based on locally linear embedding (LLE) and random forest (RF). From the received radio signal, a set of cyclic spectrum features are first calculated, and the LLE computes low dimensional, neighbourhood preserving embeddings of high dimensional data for classification. Then the detecting signal is classified by the trained random forest to test whether the primary user exists or not. Compared with SVM and PCA-SVM, the performance of our proposed algorithm is evaluated through simulations. Experimental results show that the performance of our proposed algorithm is much better than compared algorithms in low signal-to-noise ratio environments.