Optimal Max-SINR scheduling in full-duplex OFDMA cellular networks with dynamic arrivals

Abstract

In this paper, we propose an optimal algorithm for scheduling in a full duplex Orthogonal Frequency Division Multiple Access network. This network exhibits a full-duplex base station that concurrently communicates with a pair of one uplink, and one downlink half duplex user equipment (UE). Our objective is to maximize the system's sum-SINR i.e., the sum of the SINR values of all the UEs that are allocated resources. We formulate our optimal algorithm as an integer linear optimization problem. Our formulation of the problem has the originality of incorporating a non-full buffer traffic model. In order to deal with such type of traffic, we introduce a resource utilization factor along with additional constraints on the optimization. This factor helps tune the problem, affecting outputs such as UE SINR and throughput. We compare our algorithm with an iterative greedy heuristic full duplex Max-SINR algorithm we previously proposed, as well as with the related works. Simulation results show that the optimal solution provides improved throughput values in comparison with the heuristic algorithm, and that it constitutes a more realistic approach to scheduling than the proposals in the state-of-the-art. Finally, we study the effect of partial state information on the performance of our algorithm. Under imperfect channel information, our FD algorithm still provides higher UE throughput than traditional half duplex systems.