System-to-distribution parameter mapping for the Gini index detector test statistic via artificial neural networks. (July 2020)

• Record Type: Journal Article
• Authors: Lemes, Alan L.; Guimarães, Dayan A.; Masselli, Yvo M.C.
• Is Part Of: Computers & electrical engineering. Volume 85(2020)
• Publication Date: 2020-07
• Subjects: Artificial neural networks -- Binary hypothesis test -- Cognitive radio -- Cooperative spectrum sensing -- Gini index detector -- Goodness-of-fit; Computer engineering -- Periodicals; Electrical engineering -- Periodicals; Electrical engineering -- Data processing -- Periodicals; Ordinateurs -- Conception et construction -- Périodiques; Électrotechnique -- Périodiques; Électrotechnique -- Informatique -- Périodiques; Computer engineering; Electrical engineering; Electrical engineering -- Data processing; Periodicals; Electronic journals; 621.302854
• DOI: 10.1016/j.compeleceng.2020.106692
• Languages: English
• ISSNs: 0045-7906

Abstract

Highlights: The theoretical performance of some detectors for spectrum sensing are hard to find. The performance metrics demand the knowledge of the test statistic distribution. The distributions that characterize the Gini index detector are found. Artificial neural networks are used to perform system-to-distribution parameter map. Graphical abstract: Abstract: The Gini index detector (GID) was recently proposed for cooperative spectrum sensing (CSS) in cognitive radio networks. It has low computational complexity, robustness against unequal and time-varying noise and received signal powers, and can outperform state-of-the-art detectors. In this article, artificial neural networks (ANNs) are applied to map the CSS system variables into those that parameterize the probability distributions of the GID test statistic under the hypotheses of absence ( H 0 ) and presence ( H 1 ) of the primary sensed signal. The results concerning the goodness-of-fit of the GID test statistic to candidate probability distributions demonstrate that the Stable distribution adequately characterizes the statistic under H 0, whereas the Generalized Extreme Value distribution best applies to H 1 . Two ANNs are developed to establish the system-to-distribution parameter mapping, allowing theoretical calculations of the CSS performance metrics and the decision threshold via closed-form expressions. The theoretical results are verified by computer simulations.