In cognitive radio (CR) networks, spectrum sensing is an essential operation for unlicensed users to discover available spectrum opportunities for communications. However, current spectrum sensing techniques have a drawback that they can only detect the existence of signals but they cannot differentiate whether the detected signals come from licensed users or unlicensed users. This drawback leads to a significant waste of spectrum resources, which greatly limits the spectrum utilization of unlicensed users. In this paper, we investigate the issue of differentiating the licensed user and unlicensed user signals. A novel spectrum sensing scheduling scheme is proposed to solve this issue. Under the proposed spectrum sensing scheduling scheme, accurate sensing results on licensed users are obtained. Simulation results show that the proposed spectrum sensing scheduling scheme outperforms the scenario without signal differentiation in terms of higher unlicensed user throughput. To the best of our knowledge, this is the first work that investigates the signal differentiation issue between licensed users and unlicensed users in CR networks.