It was shown that most of the radio frequency spectrum was inefficiently utilized. To fully use these spectrums, cognitive radio networks have been proposed. The idea is to allow secondary users to use a spectrum if the primary user (i.e., the legitimate owner of the spectrum) is not using it. To achieve this, secondary users should constantly monitor the usage of the spectrum to avoid interference with the primary user. However, achieving a trustworthy monitoring is not easy. A malicious secondary user who wants to gain an unfair use of a spectrum can emulate the primary user, and can thus trick the other secondary users into believing that the primary user is using the spectrum when it is not. This attack is called the Primary User Emulation (PUE) attack. To prevent this attack, there should be a way to authenticate primary users' spectrum usage. We propose a method that allows primary users to add a cryptographic link signature to its signal so the spectrum usage by primary users can be authenticated. This signature is added to the signal in a transparent way, such that the receivers (who do not care about the signature) still function as usual, while the cognitive radio receivers can retrieve the signature from the signal. We describe two schemes to add a signature, one using modulation, and the other using coding. We have analyzed the performance of both schemes.