Research On Dynamic Trust Model In Opportunistic Networks

With the development of mobile communication terminal and wireless communication technology, the opportunity networks develops fast and has become one of the hot research areas of the mobile Internet. The opportunity networks has the characteristics of distributed and self-organization, so it has been widely used and has become an important part of the future Internet of things and pervasive computing. However, the openness of the networks has the risk in the aspects of route forwarding, security and privacy protection. The selfish and malicious behavior of nodes in the networks will seriously affect the performance of the whole network. In order to ensure the success rate of messages delivery ratio and reduce the messages delay, the node needs to establish the corresponding trust mechanism which is the basis of making routing decision.In this paper, we analyze the abnormal behavior of the selfish nodes and malicious nodes in opportunistic networks, then we propose a dynamic trust model based on the characteristics of the non fully connected and self-organizing networks. Selfish nodes refused to receive other nodes' messages or discarded immediately after received in order to save their energy and memory. Malicious nodes not only discard the received message and send out a lot of spam messages, but also use collusive attack to evade node monitoring. The monitoring mechanism of nodes in the dynamic trust model can record the interaction information between them and other nodes, then the node quantifies the recorded information as the direct trust between nodes, including satisfaction, health degree, connectivity, etc.. In order to measure the degree of trust between nodes more accurately, the node considers its own preference and receives the recommendation trust of other nodes selectively. The dynamic trust model aggregates the direct trust and recommendation trust of nodes in a certain way, then stored the combined trust which will update periodically into the trust list. The node whether sends a message to a node or not decided by the comprehensive trust value, that is to say, the node refuses to send a message to the node that is not trusted. For the conspiracy attacks of malicious nodes, the model counts the sender and confirmation of the feedback messages, and continue to collect recommended feedback by other nodes at the same time. The model is able to detect collusion attack nodes according to the proportional of feedback message sender. In this paper, the trust model is applied to the typical routing algorithm Prophet, Spray And Wait and Epidemic.The paper uses the open source simulation tool ONE to simulate the proposed dynamic trust model and verify the performance of the model in the opportunistic network. The result of simulation shows that the nodes can detect a large number of abnormal nodes after a certain period of trust accumulation, which may reduce the risk of transmitting messages to the abnormal nodes and receiving the spam messages.