针对瞬时故障和间歇性故障的NoC链路容错方法
发布时间：2018-07-28 14:12
【摘要】：片上网络中链路是路由器之间连接的关键通路,其发生故障将严重影响网络性能.针对这一问题,提出了一种针对瞬时和间歇性故障的高可靠链路容错方法,该方法可以在网络中实时检测数据是否发生错误,并以此定义瞬时故障和间歇性故障,从而进行容错.在减轻网络拥塞和延时的同时,保证了数据的正确传输,有效保障了系统的高可靠性.当链路中发生瞬时故障导致数据出错且不能正确纠正时,通过设置的重传缓冲区内备份的数据重新进行传输.当链路中发生间歇性故障导致数据出错且不能正确纠正时,数据包传输被截断,对被截断的数据重新添加头微片或尾微片,从而进行重新路由或资源释放.实验结果表明:该容错方法在不同故障情况下较对比对象,均较大地降低了延时,提高了吞吐率,该方法能有效地提高网络的可靠性,保证了系统性能.
[Abstract]:On-chip network link is the key path between routers, and its failure will seriously affect the performance of the network. In order to solve this problem, a highly reliable link fault tolerant method for transient and intermittent faults is proposed, which can detect data errors in real time and define transient and intermittent faults. In order to make fault tolerance. While reducing network congestion and delay, it ensures the correct transmission of data and effectively ensures the high reliability of the system. When an instantaneous fault occurs in the link which results in data error and cannot be corrected correctly, the data backed up in the set retransmission buffer is transferred again. When an intermittent fault occurs in the link which leads to data error and cannot be corrected correctly, the packet transmission is truncated, and the header or tail microchip is added to the truncated data, and then the data is rerouted or released. The experimental results show that the fault tolerant method can greatly reduce the delay and increase the throughput compared with the contrast object in different fault cases. The method can effectively improve the reliability of the network and ensure the system performance.
【作者单位】： 合肥工业大学计算机与信息学院;合肥工业大学电子科学与应用物理学院;
【基金】：国家自然科学基金项目(61474036,61274036,61371025,61574052);国家自然科学基金青年科学基金项目(61402145) 安徽省自然科学基金青年基金项目(1508085QF138);安徽省自然科学基金项目(1508085MF117)~~
【分类号】：TN47
本文编号：2150471
[Abstract]:On-chip network link is the key path between routers, and its failure will seriously affect the performance of the network. In order to solve this problem, a highly reliable link fault tolerant method for transient and intermittent faults is proposed, which can detect data errors in real time and define transient and intermittent faults. In order to make fault tolerance. While reducing network congestion and delay, it ensures the correct transmission of data and effectively ensures the high reliability of the system. When an instantaneous fault occurs in the link which results in data error and cannot be corrected correctly, the data backed up in the set retransmission buffer is transferred again. When an intermittent fault occurs in the link which leads to data error and cannot be corrected correctly, the packet transmission is truncated, and the header or tail microchip is added to the truncated data, and then the data is rerouted or released. The experimental results show that the fault tolerant method can greatly reduce the delay and increase the throughput compared with the contrast object in different fault cases. The method can effectively improve the reliability of the network and ensure the system performance.
【作者单位】： 合肥工业大学计算机与信息学院;合肥工业大学电子科学与应用物理学院;
【基金】：国家自然科学基金项目(61474036,61274036,61371025,61574052);国家自然科学基金青年科学基金项目(61402145) 安徽省自然科学基金青年基金项目(1508085QF138);安徽省自然科学基金项目(1508085MF117)~~
【分类号】：TN47
【相似文献】
相关期刊论文 前2条
1 陶海东;张华;;3例间歇性故障的检修[J];中国有线电视;2006年18期
2 ;[J];;年期
，本文编号：2150471
本文链接：https://www.wllwen.com/kejilunwen/dianzigongchenglunwen/2150471.html
