Cooperative Spectrum Sensing in Heterogeneous Cognitive Radio Networks Based on Normalized Energy Detection

Abstract

Spectrum sensing is an essential technology to detect the presence of primary users' (PUs) signals in cognitive radio (CR) networks. To achieve desirable performances under shadowing and fading environments, cooperative spectrum sensing has been proposed as a promising scheme, which takes advantage of spatial and multiuser diversity gain. In this paper, we investigate cooperative sensing (CS) in a heterogeneous CR network scenario, where each secondary user (SU) may be equipped with different numbers of receive antennas and may have different signal processing capabilities, e.g., sampling rates. By considering the discrepancy in sensing reliability of different SUs, we propose an optimal normalized energy detection-based CS (NED-CS) scheme by virtue of the principle of log-likelihood ratio test (LRT). This LRT-based NED-CS scheme also takes into account the reporting errors at a fusion center (FC), which may receive erroneous results from some SUs due to channel fading and quantization errors. Our derivation shows that the detection statistic of our proposed LRT-based NED-CS scheme is the linear combination of modified local detection statistics, and the optimal combining coefficient can be shown to be a function of the numbers of antennas, the received signal samples, the received signal-to-noise ratios (SNRs), and the variances of reporting errors at each SU. Meanwhile, to drop the prerequisite on a priori information on the received SNRs of all SUs and the variances of reporting errors and to simplify the decision making and threshold setting, a simplified LRT (SLRT)-based NED-CS scheme is proposed as well. Furthermore, we analyze and compare the performance of our proposed LRT-based NED-CS scheme with existing NED-CS schemes, including the equal gain combination (EGC) method and the maximum normalized energy (MNE) detector, with reporting errors under both additive white Gaussian noise (AWGN) and Rayleigh fading channels in a heterogeneous CR network. We derive the probability density function (PDF) of received SNR at each SU and obtain the series expansions and closed-form approximate expressions of detection and false alarm probabilities of our proposed LRT-based NED-CS scheme. Our analytical results match simulation results well, and both of them show that the proposed LRT- and SLRT-based NED-CS schemes perform significantly better than the existing EGC and MNE methods.