Confidence Skewing Problem and Its Correction Method in Mimic Arbitration Mechanism

Arbitration mechanism is the core element of Cyberspace mimic defense (CMD) scheme, providing the security effect of exception perception and exception shielding for the protected system. The existing arbitration methods confront the problem of confidence skewing caused by attack changes when calculating confidence. This paper proposes a correction method for confidence calculation based on Logistic function (Lcorrection). The method makes a grading on influence for different historical periods and filters the noise of “overheating” exception outputs so that the confidence calculation process can adapt to the attack changes in real applications. The evaluations show that the L-correction can adequately handle the confidence skewing and skewingbased confidence cheating, improving the security of mimic arbitration mechanism.