文献类型：期刊文献

英文题名：Balancing Gaussian Efficiency and Impulsive Robustness: A Weighted Rank-Energy Detector for Cognitive IoT

作者：Mou, Jie[1]; Yang, Yibang[1]; Lai, Huadong[2]; Dai, Jisheng[3]; Xu, Weichao[1]

机构：[1] Guangdong University of Technology, School of Automation, Guangzhou, 510006, China; [2] Guangdong Ocean University, School of Electronics and Information Engineering, Zhanjiang, 524088, China; [3] Donghua University, College of Information Science and Technology, Shanghai, 201620, China

年份：2026

外文期刊名：IEEE Internet of Things Journal

收录：EI(收录号：20263521394680)、Scopus(收录号：2-s2.0-105048649396)

语种：英文

外文关键词：Cognitive radio - Correlation detectors - Gaussian distribution - Gaussian noise (electronic) - Impulse noise - Internet of things - Pulse shaping circuits - Radio systems - Statistical tests - Statistics - Trellis codes

外文摘要：Achieving high detection efficiency in Gaussian environments while maintaining robustness against non-Gaussian outliers presents a trade-off in spectrum sensing for cognitive radio-enabled Internet of Things (CR-IoT) networks. To balance these objectives, we propose the Weighted Rank-Energy Detector (WRED). The design is motivated by the decomposition of the Likelihood Ratio Test (LRT) under the Gaussian assumption, which reveals a linear structure comprising an energy term and a correlation term. WRED adopts this form by retaining the Energy Detector (ED) component and substituting the outlier-sensitive linear correlation with the rank correlation coefficient, Kendall’s Tau (KT). We establish an analytical framework under the Bernoulli-Gaussian Mixture (BGM) noise model and derive the statistical moments. Subsequently, we optimize the detector by maximizing the Deflection Coefficient (DC). A practical weight implementation scheme is also developed. Numerical results show that WRED approaches the performance upper bound of the LRT in Gaussian environments while maintaining robustness comparable to rank-based methods in impulsive noise. ? 2014 IEEE.