Journal article
A. Onumanyi, E. Onwuka, A. Aibinu, O. Ugweje, M. Salami
2014
2014
Semantic Scholar
Cite
Cite
APA
Click to copy
Onumanyi, A., Onwuka, E., Aibinu, A., Ugweje, O., & Salami, M. (2014). Comparative sensitivity analysis of energy detection techniques for cognitive radio application.
Chicago/Turabian
Click to copy
Onumanyi, A., E. Onwuka, A. Aibinu, O. Ugweje, and M. Salami. “Comparative Sensitivity Analysis of Energy Detection Techniques for Cognitive Radio Application” (2014).
MLA
Click to copy
Onumanyi, A., et al. Comparative Sensitivity Analysis of Energy Detection Techniques for Cognitive Radio Application. 2014.
BibTeX Click to copy
@article{a2014a,
title = {Comparative sensitivity analysis of energy detection techniques for cognitive radio application},
year = {2014},
author = {Onumanyi, A. and Onwuka, E. and Aibinu, A. and Ugweje, O. and Salami, M.}
}
Abstract
With sensitivity being an important factor in spectrum sensing based Cognitive Radio (CR) application; it remains unclear which out of the many existing Energy Detector (ED) techniques provides the best sensitivity performance for CR application. Consequently, this paper reports a study of some known parametric and non-parametric Energy Detector (ED) schemes for Cognitive Radio (CR) application towards providing relevant information. The models studied are the Simple Periodogram (SP), Welch Periodogram (WP), Multi-Taper (MT), Yule-Walker (YW), Burg (BG), and Covariance (CV). Each technique was developed using known mathematical models and appropriate signals were simulated for comparative analysis. However, owing to the limitation of the typical Receiver Operating Characteristic (ROC) curve to infer comparative information, our study proposes a decomposition of the ROCs of each technique into respective detection and false alarm probability curves in comparison with estimated threshold levels to enhance comparative inference. From our findings, it was observed that a detection performance gain of about 50% can be achieved when using parametric techniques over nonparametric methods especially in low SNR conditions. Furthermore, a possible 15dB increase in sensitivity performance can be achieved in narrow than wideband sensing for all techniques. Finally, an increase in sensing time might not necessarily improve detection performance in low SNR conditions provided a low false alarm performance must be maintained.
