A Hypothesis Testing-based Cyber Deception Framework under Sequential Reconnaissance

Abhinav Kumar, Swastik Brahma, Charles A. Kamhoua

Abstract: In this paper, we present a novel sequential hypothesis testing framework to model an attacker’s process of gathering and processing information sequentially during the reconnaissance phase. The attacker’s optimal sequential decision rule, that enables it to promptly terminate its sequential reconnaissance process when its gathered information yields a desired decision-making quality, has been characterized. Further, employing the developed framework, we characterize optimal strategies that a defender can employ to feed the attacker with falsified information such that the expected time taken by the attacker for its malicious decision-making is maximally prolonged. Criterion that such falsification strategies must satisfy to blind the attacker and make it take indefinitely long to decide has been presented. Numerical results have been presented to provide important insights into the developed strategies.