Barrier coverage is a critical issue in wireless sensor networks deployed in security applications (e.g., border protection), whose performance strongly depends on the locations of sensor nodes. Existing works on barrier coverage typically assume that sensor nodes have accurate location information, which is not reasonable or practical for many real sensor networks. In this paper, we study the barrier coverage problem when sensor nodes have location errors and deploy mobile sensor nodes to improve barrier coverage if the network is not barrier-covered after initial deployment. We analyze the effects of location errors for barrier coverage and propose a fault-tolerant weighted barrier graph to model the barrier coverage formation problem. Based on the graph, we prove that the minimum number of mobile sensor nodes needed to achieve barrier coverage with a guarantee is the length of the shortest path on the graph. Furthermore, we improve the computational efficiency of the fault-tolerant barrier coverage formation algorithm by removing unnecessary edges on the graph. Experimental results validate the correctness of our analysis and the proposed algorithms.