Authors:
Ahmed Y. Zakariya
1
;
2
;
Sherif I. Rabia
1
;
2
and
Waheed K. Zahra
3
;
2
Affiliations:
1
Department of Engineering Mathematics and Physics, Faculty of Engineering, Alexandria University, Alexandria 21544, Egypt
;
2
Institute of Basic and Applied Sciences, Egypt-Japan University of Science and Technology (E-JUST), New Borg El-Arab City, Alexandria 21934, Egypt
;
3
Department of Engineering Physics and Mathematics, Faculty of Engineering, Tanta University, Tanta, 31111, Egypt
Keyword(s):
Cognitive Radio, Markov Decision Process, Reinforcement Learning, Ambient Backscatter.
Abstract:
In RF-powered backscatter cognitive radio networks, while the licensed channel is busy, the SU can utilize the primary user signal either to backscatter his data or to harvest energy. When the licensed channel becomes idle, the SU can use the harvested energy to actively transmit his data. However, it is crucial for the secondary user to determine the optimal action to do under the dynamic behavior of the primary users. In this paper, we formulate the decision problem as a Markov decision process in order to maximize the average throughput of the secondary user under the assumption of unknown environment parameters. A reinforcement learning algorithm is attributed to guide the secondary user in this decision process. Numerical results show that the reinforcement learning approach succeeds in providing a good approximation of the optimal value. Moreover, a comparison with the harvest-then-transmit and backscattering transmission modes is presented to investigate the superiority of the
hybrid transmission mode in different network cases.
(More)