Fault management has long been an indispensable component for controlling and managing telecommunication networks. To prevent huge data losses, it is necessary to develop a fast and efficient fault localization mechanism. In this work, we study the problem of multi-failure localization in transparent optical networks. In this context, a correlation-based approach is introduced to exploit the quality of transmission of acquired lightpaths to localize the faulty links. The proposed search-based framework can be implemented by leveraging any search algorithm. One may utilize the exhaustive search method to localize faulty links more accurately, but at the cost of taking more time. On the other hand, one may utilize intelligent search methods with the aim of reducing the required time for localization at the expense of accuracy. However, we propose to use both of the search approaches together. In this way, faulty links are first localized by the intelligent search methods to reroute and restore the failed traffic as fast as possible to prevent further loss of data. To this aim, a genetic algorithm (GA) is proposed to search among the suspected links. Subsequently, exhaustive search method can be utilized to localize failures more accurately without time constraint and then send technicians to the right site to recover the faulty links. The obtained results reveal that the proposed GA approach achieves overall high localization accuracy (98.6%–100%) that is insignificantly affected as the traffic load decreases. Dual and triple-failure incidents are localized within 42–80 ms and 596–2180 ms, respectively. It is shown that the mean time required for localizing failures using the GA search algorithm is significantly lower than exhaustive search approach by several orders of magnitude. Hence, the proposed GA-based fault localization algorithm can reduce the average time required to restore the traffic in case of failures, applicable for the restoration applications.

长期以来，故障管理一直是控制和管理电信网络不可或缺的组成部分。为防止大量数据丢失，需要开发一种快速高效的故障定位机制。在这项工作中，我们研究了透明光网络中的多故障定位问题。在这种情况下，引入了一种基于相关性的方法来利用获取的光路的传输质量来定位故障链路。所提出的基于搜索的框架可以通过利用任何搜索算法来实现。可以利用穷举搜索方法更准确地定位故障链路，但代价是花费更多时间。另一方面，人们可以利用智能搜索方法来减少定位所需的时间，但会牺牲准确性。然而，我们建议同时使用这两种搜索方法。通过这种方式，故障链路首先通过智能搜索方法进行定位，以尽可能快地重新路由和恢复故障流量，以防止数据进一步丢失。为此，提出了一种遗传算法（GA）来搜索可疑链接。随后，可以利用穷举搜索方法更准确地定位故障而不受时间限制，然后将技术人员派往正确的站点以恢复故障链路。获得的结果表明，所提出的 GA 方法实现了整体高定位精度（98.6%–100%），随着流量负载的减少，该精度受到的影响微乎其微。双重和三重故障事件分别定位在 42-80 毫秒和 596-2180 毫秒内。结果表明，使用 GA 搜索算法定位故障所需的平均时间明显低于穷举搜索方法几个数量级。因此，所提出的基于 GA 的故障定位算法可以减少故障情况下恢复流量所需的平均时间，适用于恢复应用。