A Robust Computation Model for Trust Degrees in Trusted Networks

摘要

In order to realize trust management in large scale distributed networks, quantification models of direct trust reputation, and indirect trust, are proposed respectively. The direct trust model integrates transaction satisfaction degrees with weights based on trust attenuation coefficient. Weights are set in increasing order to obtain sensitivity to new transactions. The reputation model set nodes whose direct trust evaluation obviously deviates from mean with small weights, to enhance its ability of antiattacking. The indirect trust model uses a dynamic balance coefficient to deal with risks from attacked reputation and small samples of direct transactions.