Ünal, Deniz2018-08-312018-08-312018-082018-082018-08-28http://hdl.handle.net/11693/47759Cataloged from PDF version of article.Thesis (M.S.): Bilkent University, Department of Electrical and Electronics Engineering, İhsan Doğramacı Bilkent University, 2018.Includes bibliographical references (leaves 62-65).Improving spectral efficiency is a key objective in next generation wireless networks. Recent advances in self-interference cancellation techniques made in-band full-duplex wireless communications possible. Unlike half-duplex systems which require orthogonal frequency or time resources to separate transmission and reception, in-band full-duplex radios utilize the channel bidirectionally and theoretically can double the ergodic capacity. However due to cost, power consumption and complexity constraints, mobile stations may not support this technology. In this work, operation of full-duplex base stations with legacy half-duplex mobile stations is considered. An inherent issue of this topology is the presence of signi cant inter-user interference between half-duplex mobile stations. In order to manage this at network level, an optimization problem is formulated for a cellular network topology. Solution methods and their corresponding sum throughput are compared with respect to the number of mobile stations. An analytic solution is presented to evaluate the throughput and full-duplex gains of random pairing method for the same scenario. Then the case of limited channel state information is evaluated and a learning strategy is introduced to extend the user pairing problem to a continuous case. Performance evaluation with 100 mobile stations show that the proposed learning strategy can reduce the overhead airtime more than 80%. A weighted random sequential algorithm which is integrated to the learning process is proposed, and its performance evaluation under random walk and random waypoint mobility cases are performed.xii, 65 leaves : charts ; 30 cm.Englishinfo:eu-repo/semantics/openAccessWireless NetworksFull-DuplexMobilityStation PairingThesisB158923