A modified cyclostationary spectrum sensing based on softmax regression model
Li Zhang, Hai Huang, Xiaojun Jing · 2016
In this paper, a modified Cyclostationary Feature Detection (CFD) algorithm based on softmax regression model is proposed, which targets for improving the sensing performance under low signal-to-noise ratio (SNR). The softmax regression model provides excellent classification performance; moreover, it is simple and runs fast compared with other classification algorithms in machine learning. Therefore we choose this model to be used for spectrum sensing, which happens to be a binary hypothesis-testing problem. In this method, cyclostationary characteristic parameters were extracted as the training set and test set Then the softmax classifier which had been trained was used to detect the primary user (PU) exists or not. Numerical results show that the proposed spectrum sensing method has better sensing performance than the traditional CFD method.