| dc.description.abstract | A con-man deception appears in services from cyberspace, e.g., in cloud services. A cloud-service provider deceives by repeatedly providing less service than promised and deliberately avoids service monitoring. Such a repeated shortfall is beneficial for the cloud-service provider but victimizes the service’s legitimate consumers. This deception is called a con-man deception. A con-man-resistant trust algorithm is used as a proactive measure against such deception, reducing the deception’s severity on the victim’s end. This trust algorithm detects a con-man deception by evaluating a cloud service’s expected versus actual behavior. This detection application reveals the con-man-resistant trust algorithm’s previously veiled properties. With this dissertation, a study of these properties reveals some necessary enhancements for this algorithm. The previous con-man-resistant trust-algorithm applications only considered the pattern of service-shortfall repetition. However, for cloud applications, the service-shortfall magnitude at each repetition is also important. Hence, an exponential growth-function-based extension of this algorithm is proposed and implemented. The algorithm’s initial parameter configuration has a significant influence on the con-deception detection pace. Some consumers tolerate intense repetition of service shortfall, and some consumers can tolerate mild repetition. Hence, the deception-detection pace has a correlation with the consumer’s perspective. A machine-learning extension of the con-man-resistant trust algorithm can ascertain a consumer’s perspective by analyzing that consumer’s historical usage of the same cloud service. The result of this learning is a parameter configuration that reflects the consumer’s perspective. The loss associated with a con deception is significant on the consumer’s side. Hence, the work presented in this dissertation contributes to cybersecurity by attempting to minimize such deception in cyberspace. | en_US |
