Deep learning application for sensing available spectrum for cognitive radio: An ECRNN approach

Spectrum sensing (SS) is a concept of cognitive radio systems at base transceiver stations that can find the white space i.e. licensed spectrum owned by primary users (PU), for transmission over a wireless network without any channel interference. The cognitive radio network is designed to overcome the problem of the limited radio frequency spectrum as most of the applications are dependent on wireless devices in 5G. The major concern that arises here is the detection of spectrum availability. The traditional approaches can solve this issue but consume a large amount of time and prior information about PU and spectrum. The objective of this paper is to give a solution to resolve such issues. In this paper, we have used the learning capabilities of deep learning algorithms such as Convolution neural network (CNN) and Recurrent neural network (RNN) for spectrum sensing without prior knowledge of PU. The proposed model is termed ensemble CNN and RNN (ECRNN) to learn the features of spectrum data and predict the spectrum availability at base transceiver stations in 5G. The simulation result of the ECRNN showed the improvement of accuracy of the system with a reduction in losses that occurred during the false alarm of prediction as well as an improvement in the probability of detection. ECRNN had analyzed PU statistics and result in better spectrum sensing. This paper also supported multiple SUs that would increase the speed of spectrum sensing and data transmission over the available limited spectrum at the same time.