Load Profile-based Efficiency Metrics for Code Obfuscators

Abstract

Code obfuscation techniques are gaining more attention and importance as ubiquitous computing becomes commonplace. The necessity to protect intellectual property rights as well as the need to prevent tampering with code running on all sorts of autonomous devices increases the demand for high quality code obfuscator tools. Currently, the principal quality metrics for evaluating code obfuscators mostly revolve around aspects such as security and functional correctness of the generated code. Nevertheless, power consumption plays a central role in handheld devices, as battery life and endurance is usually the bottleneck in achieving an acceptable level of user satisfaction with the system. Consequently, the criteria for choosing the right code obfuscator should be extended to also take into account the impact of obfuscation on the overall power consumption. This paper presents one viable approach for evaluating code obfuscators in regard to a power consumption level of the obfuscated code. The methodology is based upon load profiles. The performance of the solution has been tested using various commercial code obfuscators. The results show that there are significant differences in power consumption levels between original and obfuscated code. In order to select an obfuscation tool it is not enough to rely solely on non-power related attributes. High increase in a power consumption level may be totally inappropriate on mobile devices, despite the best obfuscation achieved with that particular tool. Accordingly, power consumption level should be incorporated into a set of quality metrics for code obfuscators.