Heterogeneity, dynamicity and variability that are inherited in dynamic distributed networks have come with their security challenges. First, all devices on the network must be trusted to operate according to their designated security level. Hence, any accidental or suspicious access may breach the security of the protected resources. Second, the architecture of the dynamic networks, if not designed very carefully, can make sharing appropriate information among devices difficult or impossible. In this paper, we present an approach for trust management in dynamic distributed networks. Our approach is based on establishing two defense lines to protect the resources of the dynamic distributed networks from suspicious access and obtrusive penetrating trials. For the first line, we employ a machine learning approach that serves as an outer shield which faces the penetrating trials. If this shield is penetrated, then the inner strong semantic-based shield defends against the penetrating attacks and prevents them from reaching the protected resources. We show, in this paper, how machine learning models can be combined with semantic technologies to support flexible and adaptive trust management decisions. We develop a proof-of-concept implementation and give the complexity analysis for our approach.

Keywords