抗时延敏感性跨层自适应资源分配方案

Adaptive cross-layer resource allocation scheme resisting delay sensibility

为了对抗多用户OFDM系统中用户实时业务对时延的敏感性,提出一种利用Hopfield神经网络(HNN)算法的跨层自适应资源分配方案。该方案设置用户调度优先级时同时考虑物理层的信道状态信息,及媒体接入层的用户队列状态信息和等待时间等;采用HNN算法,最大化系统容量的同时降低了平均时延和丢包率。仿真结果表明,相比于传统资源分配方案,该方案可以有效保障用户的服务质量,并提高了系统的整体性能。

In order to resist delay sensibility of real-time services, this paper proposed an adaptive cross-layer resource allocation scheme employing Hopfield neural networks （HNN） algorithm in multi-user OFDM system. In setting of the user scheduling priority, this paper considered not only the channel state information （CSI） at the physical layer, but also users＇ queue state information and waiting time at media access control （MAC） layer. By the HNN algorithm, the proposed scheme decreased the average delay and packet loss ratio, and maximizing the system capacity. Numerical results show that the scheme can efficiently guarantee the quality of service （QoS） requirements and improve the system performance compared with traditional schemes.