Cognitive radio (Cr) is one of among the most promising developments, which combines optical non-orthogonal multiple access (O-NOMA) methods to increase the fifth generation (5G) spectrum’s usage efficiency. In this work, we provide a hybrid approach based on cyclic stationary feature detection (CFD) and energy detection (ED). The probability of false alarm (PFA), on which it is based, is used to choose which spectrum sensing (SS) approach to use. When PFA is less than 0.5, the ED is chosen; otherwise, the MF is. The suggested detection method in O-NOMA is investigated using Matlab-2016 simulation with reference to several metrics, such as bit error rate (BER), signal-to-noise ratio (SNR), probability of detection (PD), and PFA.