Abstract: Network coding is to apply innovative error-correction coding techniques in the network layer to improve network performance in both wired and wireless networks. It has been theoretically shown and experimentally demonstrated that if it is properly applied, it can significantly improve end-to-end network throughput, and hence has attracted tremendous attention in the last fifteen years. Unfortunately, this technique also has some serious drawbacks. One of the major problems is its vulnerability to pollution attacks, where malicious nodes can inject corrupted packets to mess up with the decoding process. To deal with this serious problem, many schemes have been proposed in the literature, but most of them are centralized in the sense that a trusted central authority may be required. In this paper, we propose a novel distributed defense scheme based on some reputation mechanism by taking advantage of node mobility. The fundamental idea is to apply an effective reputation mechanism to locate potential malicious nodes whenever suspected polluted packets are detected. We have conducted extensive comparison studies of our proposed scheme and the existing ones, and demonstrated that the proposed scheme can achieve high successful packet delivery ratio by effectively locating and isolating the malicious nodes, even when there exist multiple malicious nodes in the network.