On countermeasures to traffic analysis attacks

We make three contributions. First, we propose Shannon's perfect secrecy theory as a foundation for developing countermeasures to traffic analysis attacks on information security systems. A system violating the perfect secrecy conditions can leak mission critical information. Second, we suggest statistical pattern recognition as a fundamental technology to test an information system's security. This technology can cover a large category of testing approaches because of statistical pattern recognition's maturity and abundant techniques. Third, researchers have proposed traffic padding as countermeasures to traffic analysis attacks. By applying the proposed information assurance testing framework, we find that constant rate traffic padding does not satisfy Shannon's perfect secrecy conditions because of its implementation mechanism. We design a variant rate traffic padding strategy as an alternative, which is validated by both theoretical analysis and empirical results.