Cryptographic techniques deal with securing information against unwanted usage, while coding techniques deals with keeping data error-free and retrieving them reliably. However, both techniques share many tools, bounds and limitations. In this thesis, several novel approaches towards improving system reliability by combining cryptographic and coding techniques in several constellations are presented. The first constellation is deploying pure cryptographic functions to improve reliability issues overshadowed in systems that previously had no
reliability-supporting coding mechanisms. Such systems could have just authenticity, secrecy and/or integrity mechanisms for security services. The second constellation deploys a mixture of both cryptographic functions and error correction codes to improve the overall system reliability.
The first contribution in this thesis, presents a new practical approach for detection and correction of execution errors for AES cipher. The source of such errors could be natural or as a result of fault injection attacks. The proposed approach is making use of the two linear mappings in the AES round structure for error control.
The second contribution is investigating the possibility and ability of deploying pure cryptographic hash functions to detect and correct a class of errors. The error correction is achieved by deploying a part of the hash bits to correct a class of selected unidirectional error class with high probability. The error correction process would degrade the authentication level in a non-significant fashion. In the third and fourth contributions, we propose algorithms to improve system correctability beyond classical limits by combining coding and cryptographic functions. The new algorithms are based mainly on the fundamentals investigated in the second contribution as mechanisms to detect and correct errors. The new algorithms are investigated in terms of collision and attacking complexity, as error correction via hash matching is similar to a successful authentication attack. The resulting performance showed achievable good error correctability, authenticity, and integrity figures.
004 Data processing and computer science 621 Applied physics