论文部分内容阅读
总结目前IPv6路由查找算法优缺点,提出了一种新的IPv6路由查找算法(IBFBP).该算法结合改进的布鲁姆过滤器(IBF)与BP神经网络,将IPv6不同长度网络ID作为IBF的输入,以关键字的特征标志创建标志库(LB)进行学习,提前判断是否发生误判.并且将位数组用counter计数数组来代替,支持可删除操作,进而进行BP神经网络学习过程.理论分析和实验结果表明:该算法比已有神经网络路由查找算法需要学习的条目数平均减少了1 500倍,还降低了误判率和搜索成本,提高了查找效率.
In this paper, we summarize the advantages and disadvantages of the current IPv6 routing lookup algorithm and propose a new IPv6 routing lookup algorithm (IBFBP). Combining the improved Bloom filter (IBF) and BP neural network, IBF Input, create a flag bank (LB) by keyword signature, and judge whether it is misjudged in advance, and replace the bit array with counter count array to support the deletable operation and then carry out the BP neural network learning process. The experimental results show that the proposed algorithm reduces the number of learning items by 1 500 times compared with the existing neural network routing lookup algorithm, reduces the false positive rate and search cost, and improves the search efficiency.