基于Hopfield神经网络的HSDPA调度算法研究计算机科学与技术专业论文.docxVIP

Abstract In recent year, with the development of interaction serviece, background service and streaming service in mobile comunication system, the demand for a better cell moblile comunication syetem, with high capacity, good comunication quality and more coverage, is inreasing. And obviasly the downlink perfermence take a important part in the whole system perfermence. To meet this need, 3GPP modified the wireless interface and MAC layer, and bring a new technology called High Speed Downlink Packet Access (HSDPA). The data streaming based scheduling stratage is a key technology in HSDPA system, because the system must meet the needs of user’s service quality, consider user’s priority, provide certain data rete and time delay and maximize the system throughput. The main context of this paper is about fast packet scheduling algorithm. In packet switch network, when several users waiting for service, the scheduler need to make a optimized scheduling decision to meet the need of QoS of each user. This is the basic scheduling problem in wired network, while it is more complecated in wireless network. The wireless channel is variant, so the scheduler must consider the channel condition to fully use the channel. The context of this paper implied through simulation. First, I introduce serveral key techonology of HSDPA. Second, the real time and none realtime scheduling scheme is studied, and make a deep research on the realtime scheduling scheme including the perfermance of M-LWDF and PFMR algorithm in const rate enviroment. Forther more, a HNN-based scheduling scheme is proposed and analysised. At last, all scheme is simulated on NS-2 and the simulation result is analysised. The simulation result shows that the Hopfield Neural Network Based scheduling scheme can fully use the wireless channel and make a optimized solution between throughput and fairness, although it is not a fairness scheduling algorithm. The novel scheduling scheme get better throughput, when meeting th