| With the rapid development of the Internet and the proliferation of mobile terminals,the de-mands for high quality wireless services are expected to grow exponentially in the future.Given the limited available spectrum,increasing the data rate and supporting huge number of users have be-come the key challenges.In this regard,in-band full-duplex(FD)has appeared as a promising tech-nology for spectral efficiency improvement.Compared with the conventional frequency-division duplex(FDD)and time-division duplex(TDD)where two separate channels or two separate time slots are used for uplink and downlink transmission,FD allows simultaneous transmission and re-ception over the same frequency band,which has the potential to double the spectral efficiency.Hence,FD has been short listed as one of the key enabling physical layer techniques for the fifth generation(5G)mobile communication systems.The major challenge of FD is how to cancel the self-interference(SI).To fully realize the spectral efficiency gain provided by the FD technique,various physical layer approaches have been proposed,including antenna separation and digital cancellation,analog cancellation.The state-of-the-art cancellation methods could achieve more than 110 dB reduction of SI,paving the way for practical implementation of the FD technique.On the other hand,network-level mechanisms such as media access control(MAC)protocols and resource allocation algorithms should be carefully designed to fully unleash the potential benefits of FD technique.Therefore,in this paper,we focus on resource allocation problems in the FD communication system and high-performance low-complexity algorithms are proposed.First,we study the joint user admission and power control problem in a single cell hy-brid FD/HD network.We present a simple criterion to measure the feasibility of pairing any two uplink and downlink users based on the average channel gains,which is then used to de-fine the admissible pairs for the optimization problem.Considering the practical case where max{K,M} ? N ? K+ M,Me propose a novel "dummy user" assisted strategy and trans-form the original optimization problem into a joint user admission and power control problem,which can be efficiently solved by the K-M algorithm.Numerical results show that the proposed algorithm achieves better performance compared with random pairing based power control algo-rithms.In the next,we investigate the user pairing and power allocation problem in a heavy loaded full-duplex small cell where the number of channels is less than the number of uplink and downlink users.With the objective of maximizing the total sum-rate,a joint power allocation and user pairing problem is formulated.To solve this problem,a novel decomposition method is proposed,which decouples the power allocation and user pairing problem into two subproblems.It is proven that the optimal solution of subproblems is the optimal solution of the original problem.Then,optimal algorithms are developed for the two subproblems by introducing the novel concept of "reverse point".Numerical results show that the proposed algorithm achieves better performance compared with the baselines of the greedy algorithm and random pairing algorithm.Finally,we consider a power allocation problem and the pairing problem among uplink users,downlink users and subchannels in a full-duplex network.With the objective of maximizing the total sum-rate,a power constrained three dimensional assignment problem(AP3)is formulated,which can be decouple into two subproblems-the power allocation problem for every uplink-downlink-subchannel triple and the AP3.To solve the latter problem,we propose an upper bound based on Lagrangian relaxation and a lower bound by heuristic algorithm for AP3.Specifically,we propose a novel iterative algorithm to update the upper bound and lower bound in each iteration.Numerical results show that the algorithms have satisfactory performance with limited iterations compared with the three-sided matching algorithm and random pairing algorithm. |
PDF Full Text Request