Study On Dynamic Resource Allocation Algorithms For Multi-user OFDM System

Orthogonal Frequency Division Multiplexing (OFDM) technique is capable of suppressing inter-symbol interference (ISI) caused by multipath propagation, confronting multipath fading and improving the efficiency of spectrum resource. It is the critical technology of the fourth generation mobile communication. Dynamic resource allocation algorithm is a kind of important technology in OFDM system. In broadband OFDM system over mobile fading channels, which have different channel capacities, it is necessary to distribute the subcarriers, information bits and the signal constellations adaptively to different users according to channel instantaneous attenuations. This paper works on dynamic allocation algorithms in OFDM system based on MA for the multi-user situation.The optimal allocation algorithm is introduced based on Lagrangian algorithm. By analyzing the impractical avhievement of the optimal algorithm in the real situation, three representative sub-optimal algorithms including MAO, multi-user greedy and BABS-ACG which are close to the optimal low boundary are introduced. Computer simulations show the performance of three algorithms. The advantages and disadvantages of algorithms are also analized.By analyzing those traditional algorithms, the genetic algorithm is used to settle the resource allocation problem. A scheme of the allocation of subcarriers is converted to an individual in the GA space. Several individuals form a population. The best allocation scheme is finally searched by the GA operations of the individuals in the population after several generations.Simple GA is apt to get a local optimal solution, so it is not likely to get the best scheme. This paper uses two kinds of adaptive GA. The first one adjusts crossover and mutation probability based on fitness of individuals. The second one changes mutation probability adaptively according to the frequency of the genes. Simulations show that compared with the algorithms in the first part, allocation schemes based on adaptive GA make the transmit power slightly lower and reduce the computation burden because of the uncomplicated GA operations.