(A) study on radio resource management in OFDMA uplink systems

Abstract

Orthogonal frequency division multiple access (OFDMA) has been regarded as a very promising technique for high rate transmission. One of the major advantages of OFDMA is a capability to combat the inter-symbol interference. OFDMA has been also shown to be robust against frequency selective fading environment. And OFDMA exploits multiuser diversity in the frequency domain as well as user domain. Thus the radio resource management takes a more important role in OFDMA systems This study mainly focuses on the radio resource management for OFDMA uplink systems. First, we investigate the system throughput maximization problem. We propose an efficient subcarriers and power allocation algorithm based on Lagrangian relaxation procedure. A greedy algorithm known to be the most efficient algorithm for this problem can provide a high quality near-optimal solution, but has the disadvantage of incurring a long computation time. As this problem should be solved in a real-time environment, computation time is a very important performance measure of algorithms. In this study, a computationally efficient algorithm that provides a nearly identical quality, near-optimal solution as the greedy algorithm but requires less than 10% of the computation time of the greedy algorithm is proposed. Then, we extend our throughput maximization problem to consider the fairness issue. In this study, to provide the minimum number of subcarriers to users is used for fairness criteria. We proposed a subcarriers and power allocation algorithm which assigns the required number of subcarriers to users. Proposed algorithm outperforms two-phase algorithm in the aspects of throughput and fairness. We also proposed a subcarrier and power allocation algorithm which guarantees QoS requirement of users. In this study, the QoS requirement is defined as the minimum transmit rates of users. Computer simulations show that proposed algorithm provides better system throughput and fairness among users. In ad...