Research On Trustworthiness Problems In Wireless Sensor Networks

For the purpose of sensing effective information of the environment, Wireless Sensor Networks (WSNs) have been widely used in people’s life, such as environment monitoring, agricultural and industrial control and military affairs. But WSNs have many weaknesses since the limitations of technology and they are easily interfered by uncertain factors from inside or outside (e.g. electromagnetic interferences, physical collisions). Especially for wireless sensor nodes, they are so vulnerable that the sensed data always cannot reflect the real world situation well after the interferences. In this way, the sensor nodes may break down and have low trustworthiness. If we want to collect accurate data information, the faulty nodes should be revised in after-deployment maintenances. But before that, the trustworthiness of these nodes should be evaluated to figure out the faulty ones.Based on research, we found that some solutions have been proposed to evaluate the trustworthiness of WSNs. Although these solutions can help us to do some research on the trustworthiness of WSNs, they have many shortcomings. Firstly, most of these solutions are proposed based on the communications between sensor nodes and the behaviors of them, so many state information of WSNs need to be transmitted and this will overload the network. Secondly, the algorithms of these solutions need to be run in wireless sensor nodes, and this will consume many energy and storage resources. Also, current research couldn’t locate the faulty nodes accurately and it will be difficult for workers to revise these nodes.In order to deal with the above shortcomings, we propose the novel trustworthiness evaluation method and algorithm for wireless sensor networks in which the nodes can sense multidimensional data. The details can be described as follows:(1) For sensor nodes which can sense multidimensional data, we design the trustworthiness evaluation method based on D-S evidence theory. This method can be operated on a base station. Different dimensions of a sensor node are regarded as its different trustworthiness attributes. The trustworthiness of each attribute is evaluated firstly based on evidence theory, and then the lower and upper limits of trust degree for this node are calculated by fusing the evaluation results of different attributes. The implementation of this method is mainly based on the data sensed by nodes. It is not necessary to transmit other state information and the algorithm can be run without the participation of sensor nodes. So this method will not overload the networks. It can also save much energy and storage resources.(2) In complicated environments, the low trustworthiness of wireless sensor nodes may be produced by inside factors, also it may be produced by regional uncertain factors. For the latter, we should take measures to eliminate these factors instead of just revising the nodes. But firstly we need to figure out the regional uncertain factors exist or not and this can be given by the trustworthiness of the local region. So we propose a regional trustworthiness evaluation method based on D-S evidence theory. And the trust degree of a local region is given by fusing the judgments of deployed sensor nodes according to the combination rules of evidence theory. This method can not only save energy and storage resources, but also help people to take pertinence actions to ensure the system run normally.Additionally, extensive experiments based on actual data samples are conducted to evaluate the performance of our method. The theoretical analysis and experimental results show that our method can give effective trustworthiness evaluation for one single sensor node or a local region. Also, robustness and stability of this method are verified in the experiments.