Reputation-based Trust Update in Network Environment

In a network environment, trust management is an important security issue. A weighted sum is prevalent in trust formalization. Trust is contextual, continuous and changeful. There are many factors that can affect trust update in the whole lifetime of trust. However, these factors have not been studied thoroughly. In this paper, we propose a weighted trust formula and present an integrated trust update method. First, we will compare several recommendation trust formula and analyze the factors that affect trust update. We propose to use abnormal trust value sequence and statistical analysis to detect malicious recommenders and false recommendation trust values in the whole lifetime of trust. Assuming that direct trust weight and recommendation trust weight are independent from each other, we propose a real-time weight update method through trust value sequences. Simulation results show that our method could effectively detect malicious recommenders and malicious recommendation trust values and make trust value sequence smoother.