Because nodes in the loT(Internet of Things) are widely distributed and most of them are unattended, so single node is vulnerable to be attacked and destroyed. Once one node is destroyed or attacked, the whole loT will be faced with tremendous security problems. This paper proposed a Trust Evaluation Model of loT based on AHP and Backward Cloud Model. In the model, in order to be solve the problem that nodes in loT have limited resources, the paper made use of the improved AHP(Analytic Hierarchy Process) and Reverse Cloud Model method, which can work with simple calculation and few resources. In the last, we simulate the method using omnet++ and simulation result show that the model can evaluate the trust value of the malicious node that executes bad communication behaviours fast and also reduce the possibility of wrong judgment.