A QoS-Aware Resource Allocaiton Scheme With Limited Feedback In Downlink OFDMA Systems

In future broadband cellular systems, such as IEEE802.16 system or 3GPP Long Term Evolution (LTE) system, Orthogonal Frequency Division Multiplexing Access (OFDMA) is a preferred multiple access scheme for providing high spectral efficiency and exploiting multiuser diversity. However, the performance of OFDMA systems largely depends on the flexibility and efficiency of the frequency scheduling algorithm. As the development of wireless communication technology, a variety of new services, such as HTTP (Hyper Text Transfer Protocol) web browsing, file transfer, video streaming service and VOIP (Voice over Internet Protocol) emerged. These services require different QoS (Quality of Service) parameters such as bandwidth, packet delay, minimum transmission rate, packet loss rate, bit error rate and so on. Therefore, how to guarantee the QoS requirements of various services has become a key problem in designing resource allocation algorithm.In this paper, we modeled four different types of services which are real time voice service, real time video service, non real time data service and best effort service. Also, according to the state of queues in the base station and the service requested by the user, we defined a parameter named as priority value used for resource scheduling. Based on all of these work, a new resource allocation scheme was proposed to guarantee the QoS requirements of various services. Simulation results showed that the proposed allocation algorithm can fully sustain user's requirements under reasonable traffic loads. Meanwhile in terms of system throughput, it performed better than conventional algorithm.Besides, we also modified the feedback algorithm which is used in OFDMA systems currently. A new feedback scheme is proposed which adapts the quantity of feedback to the kind of service that users request. If user requests for real time service, the parameter M of feedback is constant. While user requests for non real time service or best effort service, its parameter M of feedback decreases as the system load increase. We evaluated the performance of this scheme by simulation. And the results proved that it has much better performance than traditional TOP-M scheme.