Multidimensional Trust Evaluation Model Based on Redemption and Data Aggregation

Traditional trust evaluation models in wireless sensor networks separate detection and evaluation from data aggregation. Data accuracy is ignored in these models. Algorithms like WTE alleviate some of the problem, but not all. For example, it cannot resist attacks from nodes with a good reputation. Another ignored problem is the false detection caused by occasional abnormal behavior, hash environment and model flaw. To solve the problems mentioned above, this paper proposes a multidimensional trust evaluation model based on redemption and data aggregation. One malicious node can launch various attacks. Bad data attack is one of them. When data are aggregated, Spatial and temporal correlation are taken into consideration. Nodes recognized to transmit bad data are refused to join the data aggregation. Isolated nodes are given a chance to return to networks if certain conditions are satisfied. But the network has less tolerance to them. The simulation prevents that the model has high detection rate and more accurate data.