Thanks to recent advances in WDM technologies, an optical fiber is capable to carry up to 200 wavelengths operating at 40 Gbps each. In such high speed networks, service disruptions caused by network failures (e.g., fiber cut, amplifier dysfunction) may lead to high data losses. A network operator should be able to promptly locate such failures, in order to perform fast restoration. Hence, an efficient fault detection and localization mechanism is mandatory for reliable network design. In previous work, we have introduced the concept of monitoring-trees (m-trees) to achieve fast link failure detection and localization. We have proposed an integer linear programming (ILP) approach for the design of an m-tree solution that minimizes the number of required optical monitors, while achieving unambiguous failure detection and localization. In this paper, we propose a novel approach, based on the well known simulated annealing meta-heuristic, for the m-tree design in WDM networks. Simulations conducted in this study show the same results as the ILP approach at much lower computation time. Our proposal can thus be applied to large-sized and very large-sized networks.