No abstract

With an exponential growth in smartphone applications targeting useful services such as banks, healthcare, m-commerce, security has become a primary concern. The applications downloaded from unofficial sources pose a security threat as they lack mechanisms for validation of the applications. The malware infected applications may lead to several threats such as leaking user's private information, enforcing malicious deductions for sending premium SMS, getting root privilege of the android system and so on. Existing anti-viruses depend on signature databases that need to be updated from time to time and are unable to detect zero-day malware. The Android Operating system allows inter-application communication through the use of component reuse by using intents. Unfortunately, message passing is also an application attack surface. A hybrid method for android malware detection by analysing the permissions and intent-filters of the manifest files of the applications is presented. A malware detection framework is developed based on machine learning algorithms and on the basis of the decision tree obtained from ID3 and J48 classifiers available in WEKA. Both algorithms gave same results with an error percentage of 6 per cent. The system improves detection of zero day malware.

The 8-queens problem of placing 8 non-attacking queens on an 8x8 chessboard is used to hide message in an image. The method helps in randomizing the bit selection in a cover image for hiding purpose. Cover image is divided into blocks of 8x1 bytes and then masked with solutions of the 8-queens problem. Bits from the block are collected corresponding to the 8-queen solution to make a 7 bit string. LSB of the block is not considered. It gives a number in the range of 0 to 127. If a bit string, corresponding to the 8-queens solutions, matches with ASCII code of the first character from message, the corresponding solution number of the 8-queens problem is encrypted using RC4, and the cipher is stored in first block of the cover. This encrypted value works as key. The solution number corresponding to next character is XORED with the key and the resultant value is embedded in the LSB of next block. The algorithm has been tested with cover of different image file formats like BMP, PNG and TIFF. The algorithm provides very good capacity, imperceptibility and robustness.

No abstract

No abstract