Multiple-Antenna Cooperative Spectrum Sensing Based on the Wavelet Transform and Gaussian Mixture Model

Abstract

Spectrum sensing is a core technology in cognitive radio (CR) systems. In this paper, a multiple-antenna cooperative spectrum sensor based on the wavelet transform and Gaussian mixture model (MAWG) is proposed. Compared with traditional methods, the MAWG method avoids the derivation of the threshold and improves the performance of single secondary user (SU) spectrum sensing in cases of channel loss and hidden terminal. The MAWG method reduces the noise of the signal which collected by the multiple-antenna SUs through the wavelet transform. Then, the fusion center (FC) extracts the statistical features from the signals that are pre-processed by the wavelet transform. To extract the statistical features, an sensing data fusion method is proposed. The MAWG method divides all SUs that are involved in the cooperative spectrum sensing into two clusters and extracts a two-dimensional feature vector. In order to avoid complicated decision threshold derivation, the Gaussian mixture model (GMM) is used to train a classifier for spectrum sensing according to these two-dimensional feature vectors. Simulation experiments are performed in the κ - μ channel model. The simulation shows that the MAWG can effectively improve spectrum sensing performance under the κ - μ channel model.

Machine learning for cooperative spectrum sensing and sharing: A survey

“…SS performance is affected due to fading, shadowing, noise, and other interfering factors, reducing the impact of these factors, multiple antennae equipped SUs-based model is proposed under the k − 𝜇 fading channel. 61 Signals received by SUs are preprocessed by wavelet transform to remove noise from signals, and two-dimensional input features are formulated with the combination of difference between maximum and minimum eigenvalue (DMM) eigenvalue and ratio of maximum eigenvalue to matrix trace (RMET) eigenvalue. GMM technique is used for the classification of these signal features.…”

“…In Reference 57, the received signals are preprocessed by Null Space Pursuit algorithm and IQ decomposition to extract the two-dimensional features, and eigenvalues of covariance matrices are considered as a feature vector. Zhang et al 61 applied a preprocessing technique based on wavelet transform to remove noise from sensed signals and computes a two-dimensional input features composed of DMM, RMET. The features derived from eigenvalues are more robust to noise.…”

“…Further, multiple antennas improve spatial diversity and spatial multiplexing gain and hence enhance the SS performance. InReferences 39,43,46,60,61, the CSS problem is solved using multiple antennae equipped SU; however, in the rest of the ML-based methods reported in the reviewed papers, single antenna-based SUs have been used for SS.…”

A Low SNR Signal-Sensing Method Based on a Multiple-Antenna Covariance Matrix

“…Eigenvalue-based methods use the eigenvalues of the sample covariance matrix to detect signals, including the maximum-minimum eigenvalue (MME) detector [20], the energy-to-minimum eigenvalue (EME) detector [21], and the arithmetic-to-geometric-mean (AGM) detector [22]. Waveletbased methods use wavelet transform to decompose a wide frequency band into elementary building blocks of sub-bands [23,24], which leads to a high computational complexity. Covariance-based methods (CBMs) use various statistics of the sample covariance matrix to detect signals [25][26][27].…”

Effective Usage of Available Spectrum Using Dispersion Detection Technique

“…this method provides better results than energy detection by calculating different parameters for different number of antennas. [6] LRT is an optimal method but its quite complex and it requires channel information beforehand. [5]…”