Performance of cognitive radio networks under ON/OFF and Poisson primary arrival models

• Published in: 2011 IEEE 22nd International Symposium on Personal, Indoor and Mobile Radio Communications pp. 609 - 613
• Main Authors: Lirio Castellanos-Lopez, S., Cruz-Perez, F. A., Hernandez-Valdez, G.
• Format: Conference Proceeding
• Language: English
• Published: IEEE 01.09.2011
• Subjects: Analytical models; Cognitive radio; Cognitive radio networks; Mathematical model; Numerical models; ON/OFF processes arrivals; Poisson process arrivals; Probability distribution; System performance; teletraffic analysis; White spaces
• ISBN: 9781457713460; 1457713462
• ISSN: 2166-9570
• DOI: 10.1109/PIMRC.2011.6140034

In this paper, the performance of cognitive radio networks (CRNs) considering either Poisson or ON/OFF models for the primary users' arrivals is evaluated and compared. These primary arrival models have been commonly used for the teletraffic analysis of CRNs in the literature. However, the performance of CRNs under these arrivals models has not been compared in the open literature yet. This is the subject under discussion in this work. In order to provide a fair basis of comparison for the ON/OFF and Poisson arrival models, similar average primary channel occupancy characteristics are considered. Specifically, performance comparison is carried out considering both the same resource utilization factor and the same mean white space duration in both arrival models. In addition, different traffic scenarios (i.e., probability distributions of the channel holding time for primary users) are considered to provide deep insights into the performance behavior of CRNs under these arrival models. Our numerical results show the relevance of the adequate characterization of the primary channel occupancy to accurately evaluate the performance of CRNs. In particular, system performance results obtained with the Poisson model are more pessimistic than those obtained with the ON/OFF model. This is due to the fact that the variance of the white space duration is greater in the Poisson model. Finally, it is observed that performance metrics (i.e., call blocking and forced termination probabilities) for the CRN are practically insensitive to the distribution type of the channel holding time in the primary network.