【摘要】 链路故障作为一类常见的网络问题，会对网络运行的可靠性产生严重影响。然而现有的被动式方法需要较长时延来实现链路故障恢复，而传统的主动式方法则需要消耗大量且昂贵的三态内容寻址存储器(TCAM)。针对这些挑战，本文提出一种基于FFT的长短周期嵌套方法(Long and Short Nested Cycles, LSNC),用于链路故障的快速恢复。该方法在外部周期T中，利用DLBS算法对链路进行备份；每个外部周期T分为多个内部周期t,在内部周期t内进行流量监测，对外部周期建立的备份路径进行更新。LSNC利用快速傅里叶变换(FFT)获取不同区域网络流量的周期，作为外部周期T和内部周期t。本文基于GEANT数据集展开实验，证明LSNC在平均带宽利用率和备份路径长度两方面优于现有的BF和CR等较为成熟的方法。更多还原

【Abstract】 Link failure, as a common network issue, can significantly impact the reliability of network operation. However, existing passive methods typically entail longer delays for link failure recovery, while traditional active methods require substantial and costly Ternary Content Addressable Memory(TCAM). To address these challenges, this paper proposes a method called Long and Short Nested Cycles(LSNC) based on Fast Fourier Transform(FFT) for rapid link failure recovery. In this method, within an external period T, links are backed up using the DLBS algorithm. Each external period T is divided into multiple internal periods t, during which traffic monitoring takes place to update the backup paths established during the external period. LSNC utilizes FFT to obtain the periods of different region network traffic, serving as the external period T and internal period t. Experimental validation conducted on the GEANT dataset demonstrates that LSNC outperforms existing mature methods such as BF and CR in terms of average bandwidth utilization and backup path length.更多还原