The Research And Implementation Of File Password Recover System

With the development of encryption algorithm and the enhancement of people’sconsciousness about information security, computer users are used to encrypting theirprivacy data and sensitive documents. As a result, we need pay more when we forget thepassword. On the other hand, technology cannot choose users on its own initiative, somecriminals are using technology to hide their criminous clue when general users benefitfrom the convenience of technology.There is adequately practical significance of file password recovery system, but nomatter in the sphere of learning or engineering, the research and practice of file passwordrecovery is wanting at present. One project of national security department provided thebackdrop for this thesis, the work in this thesis centred on the following items：1. We did some analysis on the present situation of domestic and foreign researchin the field of file password recovery, and confirmed the significance of our research.2. We analyzed the structure of popular electronic documents, which are Microsoftoffice files, WinRar files, Adobe PDF files and ZIP files. And we emphasized on thesecurity part in the structure.3. As the most popular electronic documents are using open encryption algorithms,the speed of recovering a lost password depends on the length of key and the speed ofcomputing. When we found that it’s difficult to recovery a password with CPU, we needto do some analysis of encryption algorithm, then we find that almost all of the major fileencryption algorithm is carried out in many matrix operations, which just fits GPU.4. If we want to improve the speed of password recovering further more, anintuitive idea is to increase the number of GPU, however, we can’t integrate too manyGPUs on one motherboard because of the limitation of hardware. At this point, the theoryof distributed computing pointed out a road for us, using the distributed computing theory,we can divide the task to multiple machines in parallel computing, and each cell can usecomputer network for communication and coordination.5. Two schemes are provided to different users due to their diverse behaviors.6. We did some test of the system function and performance, Experimental result shows that GPU can improve the speed of recovering for more than one hundred timescompared with CPU, and the speed of distributed system is proportional to the number ofcomputing cell.