Effective capacity optimization for cognitive radio networks under primary QoS provisioning

Funding Number

NPRP 09-1168-2-455

Author's Department

Electronics & Communications Engineering Department

Find in your Library

https://ieeexplore.ieee.org/document/7707360

All Authors

Adel M Elmahdy; Amr El-Keyi; Tamer ElBatt; Karim G Seddik

Document Type

Research Article

Publication Title

IEEE Transactions on Communications

Publication Date

12-31-2019

doi

10.1109/TCOMM.2016.2621744

Abstract

Cognitive radios have emerged as a key enabler for opportunistic spectrum access, in order to tackle the wireless spectrum scarcity and under utilization problems over the past two decades. In this paper, we aim to enhance the secondary user (SU) performance while maintaining the desired average packet delay for the primary user (PU). In particular, we investigate the trade-off between delay-constrained primary and secondary users in cog- nitive radio systems. In the first part of this work, we use the hard-sensing scheme to make a decision on the PU activity and maximize the SU effective capacity subject to an average PU delay constraint. Second, we propose a soft-sensing scheme by dividing the PU energy interval where the PU is decided to be idle into multiple decision. We also maximize the SU effective capacity subject to an average primary user delay constraint; then, we present three modifications for the proposed soft-sensing scheme to allow for low complexity implementation that is comparable to the complexity of the hard-sensing scheme, but with better performance. The numerical results reveal interesting insights comparing our soft sensing to the hard-sensing models in terms of the optimal performance obtained from our optimization solution compared to the unconstrained PU delay baseline system studied earlier in the literature. For instance, the hard sensing system in Akin and Gursoy (IEEE Trans Wirel Commun 9(11):3354–3364, 2010) and Abdel-Malek et al. (CrownCom 156:30–42, 2015) yields a SU effective capacity of only 50 % of the ideal, perfect sensing system. On the other hand, we show that the soft sensing system yields almost 87 % of the perfect sensing performance (at a primary user arrival rate of λp= 0.1), which further increases for a larger number of decision sub-intervals.

First Page

1451

Last Page

1463

This document is currently not available here.

Share

COinS