Radio Signal Smart Deception Based on Adversarial Learning

In recent years, electromagnetic spectrum becomes more and more congested and the task of spectrum sensing meets various challenges. On the other hand, there is also a counter-demand that the transmitted signal should not be detected, intercepted, or identified by the intelligent spectrum sensing equipment. In this paper, a smart deception method for radio signal using adversarial learning under the black-box assumption is investigated. By generating smart deception, the accuracy of the intelligent electromagnetic spectrum equipment to identify the radio signal decreasing more than 50.63%. The radio signal camouflage is achieved by smart deception, which will lower the interception and identification possibility of the transmitted signal.