Discrete bit loading and power allocation for OFDMA downlink with minimum rate guarantee for users

Performance optimization of the communication networks with quality of service (QoS) considerations for all active users is of great practical importance in highly delay-sensitive applications like multipoint videoconferences and interactive video games. Using minimum data rate as the primary criterion for the QoS and the BER as a second one, in this paper, we provide the optimum discrete bit loading for an OFDMA downlink in which the QoS of all users are satisfied. We formulate the problem and solve the dual problem efficiently and without any suboptimality. Albeit the problem is non-convex, it has been shown that the duality gap tends zero in practical scenarios. We propose a novel approach to solve the dual problem by decomposing dual function and finding discrete rate allocation solution for practical applications. Also, we present a novel fast converging iterative algorithm to find the solution with much less complexity compared to that of well-known ellipsoid optimization algorithm. We analyze the complexity of these solutions, and using simulation results, show the superiority of our novel approach.