基于ATM交换结构的Hopfield神经网络调度算法

针对ATM交换结构，采用输入缓冲和每条入线在同一个时隙内可传送多于一个信元的策略，利用神经网络具有的实时性、高度并行处理能力和易于电路或光电技术实现等特点，提出了一种Hopfield神经网络调度算法。实验仿真比较表明，该方法不但大大提高了吞吐率，消除了队头阻塞造成的性能恶化，而且降低了信元丢失率和较大程度地降低了平均信元时延，提高了ATM交换结构的性能，实现了信元的优化调度。

Cell Schedule Algorithm Based on Hopfield Neural Network Model for ATM Switching Fabrics

The asynchronous transfer mode (ATM) is the choice of transport mode for broadband integrated service digital networks (B-ISDN’s). It represents the future development of networks and communication technique. A cell schedule algorithm based on Hopfield neural network (HNN) model for ATM switching fabrics (ASF) is proposed in this paper. A new energy function of HNN is employed based on dedicated input buffered cooperating with the policy of more than one cell transferred in each input line during every time slot. Experimental simulation results show that, compared with the method presented in reference 6, the approach not only improves greatly the throughput and eliminates the performance reduction due to the head of line blocking (HOL blocking), but also lowers down the cell loss probability and reduces the average latency, i.e. the performances of ASF are quite improved. It means that the optimization scheduling of the cell can be efficiently implemented by the cell schedule algorithm.