LMDGW: a novel matrix based dynamic graph watermark

摘要

Android platform induces an open application development framework to attract more developers and promote larger market occupations. However, the open architecture also makes it easier to reverse engineering, which results in the property loss for both developers and companies, and increases the risks of mobile malicious code. Therefore, copyright protection for android application is of significant importance. A class of copyright protection methods are based on Android watermark. Different from static watermark, the process of embedding and extracting of dynamic graph watermark (DGW) is based on the function path and operation process of the app, which has better concealment. In this paper, we proposed a late-model dynamic watermark based on matrix, called LMDGW. This method is proposed to overcome the shortcoming of unintuitive and vulnerable property of traditional numberal DGW. We encode a matrix with low rank into a watermark graph, and embed the graph construction statements into smali code. With the containing of sensitive block characteristics, LMDGW is able to perceive and locate the changes in the specific block. Besides, LWDGW has great performance in tamper-proof attacks. Experiment results and analysis justified that LMDGW has great data rate and robust, and is available in sensitive code locating. LMDGW is proved to be an intelligent watermarking scheme, and is enlightening for intelligent security.