This poster deals with botnets, the most dangerous kind of mobile malware, and their detection using neural networks. Unlike common mobile malware, botnets often have a complicated pattern of behavior because they are not managed by predictable algorithms but they are controlled by humans via command and control servers (C&C servers) or via peer-to-peer networks. However, they have certain common features which have been revealed by analysis of contemporary mobile botnets. These features have been used for creation of a neural network training set. Finally, the design of parallel architecture using neural network for useful detection of mobile botnets has been described.
This paper is focused to the designing, implementing and benchmarking the data cluster based on the CEPH [1] technology. The main point of this paper is aimed to the implementation with geographically separated nodes. These nodes, store points of distributed data cluster, is in different location and on different subnets. Realized test show the dependency of number of concurrent clients connected to the cluster and cluster read/write bandwidth. These test show the potential limits of the developed solution.
1 January 2013 in Advances in Intelligent Systems and Computing
This paper is about a proposal of the algorithm to derive encryption keys based on key-files using the hash algorithm SHA-3. For our upcoming design of block cipher with special key length, which is 111 bytes, we are facing the problem how to derive encryption keys, even there are for example scrypt or Argon2 key-derivation function, we would like to design algorithm specialized for generating hashes of the required length. Thus, we proposed whole new algorithm. We also wanted to forge encryption keys based on key-files and do not limit only for passwords as the source. So, this paper presents the design of the final algorithm and results from testing it. This work is also the first proposal and in the future is planned to test it more.