Abstract: Spectrum Sensing Data Falsification (SSDF) attack is one of the most important threats to the spectrum sensing for wireless cognitive radio networks. On the basis that the wireless signal in cognitive radio network is inherently sparse in frequency domain, this paper develops a distributed compressed wideband spectrum sensing approach which combines compressed sensing and average consensus algorithm and defensive against SSDF attacks. To distinguish the potential malicious node more precisely, we evaluate reputation values for each of the CR nodes which will be used at the fusion stage. At sensing stage, compressed sensing is performed at each CR nodes to sample the received wideband signal at practical complexity and cost, and then locally reconstruct the frequency domain signal. At fusion stage, the local spectrum sensing results of each CRs are fused distributed and exclude the influence of potential malicious node at the same time without a fusion center. Simulation results show that spectrum sensing performance is enhanced using our proposed model and can defend against SSDF attacks.