In this paper, we consider the problem of dynamic lightpath protection in survivable WDM networks with single link failure model. Our work concerns in how to dynamically determine a protection cycle to establish a dependable lightpath with fault-tolerance requirement. This problem is identified as NP-complete, thus a heuristic for finding near optimal solution with reasonable computation time is usually preferred. Inspired from the principle of ant colony optimization, we develop in this paper an ant-based mobile agents algorithm for this problem with improved blocking performance. To enable the new ant-based algorithm, we propose to use on each network node both a routing table that contains a set of feasible cycles between source destination nodes and also a pheromone table for mobile agents. By keeping a suitable number of mobile agents in a network to continually and proactively update the cycles in a routing table based on the current network congestion state, the routing solution of a connection request can be obtained based on simply a local searching with a reasonable computation time. Extensive simulation results upon the ns-2 network simulator show that our new algorithm can achieve a significantly lower blocking probability than the promising algorithm for dynamic lightpath protection proposed in [11] with a comparable computation complexity.