摘要
输入缓冲ＡＴＭ交换结构是实现复杂度最低的ＡＴＭ交换结构（ＡＳＦ），但由于队头阻塞问题大大限制了其应用。文中给出了基于 Ｈｏｐｆｉｅｌｄ神经网络（ ＨＮＮ）调度信元的输入缓冲ＡＳＦ方案，该方案消除了队头阻塞造成的输入缓冲ＡＳＦ性能恶化。计算机仿真表明，该方案非常有效，明显改善了输入缓冲ＡＳＦ的性能。
Of all ATM switching fabrics (ASF), input - buffering ASF is of the least im- plementation complexity. But its application is limited because of head - of - line (HOL) blocking. In this paper, we propose an input - buffering ASF based on the Hopfield neural network (HNN) scheduling cell. This scheme eliminates degeneration of performances due to HOL blocking. Sim- ulation results show the effectiveness of this method.
出处
《通信技术》
2000年第2期33-36,共4页
Communications Technology
基金
国家自然科学基金!69772027
国家教委跨世纪人才基金
关键词
ATM交换结构
调度信元
神经网络
ATM switching fabric, bypass queue, Hopfield neural network
