Optimal Efficient Discrete Resource Allocation for OFDMA Systems

This paper proposes a flexible, efficient approach to discrete rate and power allocation for OFDMA systems. We consider the weighted-sum-rate maximization problem, and utilize the Lagrange dual decomposition (as done in some recent works) along with our previous single-user approach to simplify the non-convex, combinatorial optimization problem. The result is an optimal algorithm (up to a very small worst-case duality gap) with very low complexity. Furthermore, the approach is very flexible and can easily incorporate such factors as various QoS measures, fading channels, imperfect channel knowledge and prediction errors, and error control codes provided that a simple power scalability property is satisfied.