现代电子技术Issue(5)：14-18,23,6.DOI:10.16652/j.issn.1004-373x.2016.05.004
基于二进制退避算法和流量感知的片上网络在线故障检测
Online fault detection for NoC based on binary exponential back-off algorithm and flow sensing
摘要
Abstract
The interconnecting line in network⁃on⁃chip (NoC) is liable to generate crosstalk fault,so an online fault diagno⁃sis scheme for NoC crosstalk based on flow sensing and binary exponential back⁃off algorithm (BEBOA) is proposed. The channel detector is used for channel detection to minimize the intrusion throughput of NoC when NoC channel is idle. The detection inter⁃val time parameter based on BEBOA is used to dynamically set the detection time to fast recover the network when a temporary failure is recovered. A priority arbitration mechanism was constructed to avoid the channel access conflict between application program and detection procedure. The experimental results show this scheme can detect and diagnose the fault accurately,recover the throughput quickly,and satisfy the requirement of smallest extra chip area occupation.关键词
故障检测/片上网络/二进制退避算法/互连线串扰/监控模块/优先级仲裁机制Key words
fault detection/NoC/binary exponential back-off algorithm/interconnecting line crosstalk/monitoring module/priority arbitration mechanism分类
信息技术与安全科学引用本文复制引用
崔艳,王刘涛..基于二进制退避算法和流量感知的片上网络在线故障检测[J].现代电子技术,2016,(5):14-18,23,6.基金项目
河南省高等学校重点科研计划项目（15A520120）；河南省科技厅科技攻关项目 （）
