Computation and analysis of node intending trust in WSNs

The resource constraints of wireless sensor network make it easy to attack and hard to protect. Although carefully designed cryptography and authentication help to make WSN securer, they are not good at dealing with compromised node and ageing node, whose misbehavior may impair the function of WSN. Behavior-based trust mechanism is considered as the effective supplement of traditional cryptography security, it has been widely studied in peer-to-peer network, the grid, and the pervasive computing environment. The evaluation of node trust is based on the past behavior evidence, but we hope to know the node future trust, so node trust prediction is important and significant for wireless sensor networks application such as routing, data fusion. The paper discusses how to based on past node trust to predict future node trust, including Bayesian Network modeling, prediction grading, computation of prior probabilities of node trust, computation of prior probabilities of node attribute and prediction of future node trust. In order to meet needs of different prediction of node trust for different purpose, we also discuss that how to get and store node past assorted behavior statistical data. Finally, discuss how to predict node trust in real system and how to control routing, data fusion based on prediction result.