Superimposed training-based detector for spectrum sensing in cognitive radio

Abstract

The spectrum sensing function allows a cognitive radio to determine the absence/presence of primary users' (PUs) signals in a frequency band of interest. These signals might exhibit very low-power at cognitive (or secondary) users' receivers. Thus requiring detection algorithms that work well in the very low signal-to-noise ratio (SNR) region. It is known that secondary users (SUs) can improve its detection performance if some known information about PUs' signals is available. In this regard, some PUs can use superimposed training (ST) technique for channel estimation and synchronization purposes in their own networks. However, to the authors knowledge, the exploitation of this ST information by SUs in the context of cognitive radio has not been studied yet. In this paper, a new spectrum sensing algorithm for superimposed trained PUs' signals is designed based on the Neyman-Pearson criterion. The proposed algorithm takes advantage of the ST sequence to improve the detection performance of SUs. Results show that, even with a small training-to-information power ratio, the superimposed training-based detector (STD) significantly outperforms the energy detector; specifically in the very low SNR, which is of interest in cognitive radio. Moreover, if ST is not used, the proposed STD reduces to the energy detector.