Software obfuscation was proposed as a technique to mitigate the problem of malicious code tampering, by making code more difficult to understand and consequently more difficult to alter.
In particular, "residue number coding" encodes program variables to hide their actual values, while supporting operations in the encoded domain. Some computations on encoded variables can proceed without the need to decode them back in the clear. Despite the obvious benefits of this approach, to the best of our knowledge, no implementation is available. In this paper, we describe our implementation of data obfuscation based on residue number coding. Moreover, we present an assessment of this obfuscation scheme in terms of performance overhead, when more and more program variables are subject to obfuscation. PDF version of the paper.