混合ATM信元容错识别方法  

Fault-Tolerant Recongnition Method for Mixed ATM Cells

作　　者：于鹏[1] 王晓梅[1] 于宏毅[1] 陈彦[1] 范亮[1] 

机构地区：[1]信息工程大学,河南郑州450001

出　　处：《信息工程大学学报》2014年第6期718-722,共5页Journal of Information Engineering University

摘　　要：ATM信元的识别是ATM通信中的重要环节。在无线ATM通信中,由于通信链路的不可靠性,信元在传输过程中可能会由于出现误码而导致无法识别。针对此问题,文章提出一种基于中心距离提取法和误差反向传播型神经网络相结合的方法,提取ATM信元净荷的结构特征,对含错ATM信元进行非线性容错识别。仿真结果表明,该方法对于含噪ATM信元的识别问题较常规方法有更理想的效果。The recognition of ATM cells is an important part of ATM communication. In wireless ATM environment, due to the unreliability of communication links, the type of the ATM cells may not be recognized if errors occur. In this paper, a non-linear recognition method which combines ＂center distance＂ and BP net together is proposed and applied to factual data. The simulation re- sults show that the method proposed performs much better than the regular methods.

关 键 词：BP神经网 容错识别 中心距离 

分 类 号：TP915.2[自动化与计算机技术]