Reputation System Security Research For Wireless Sensor Network

Wireless sensor network is a highly adaptive self-organizing network, a large number of inexpensive nodes are densely deployed in the monitoring area, monitor the changes of environmental parameters timely, then send this information to mankind. If the Internet has changed the way of communication between human beings, then the wireless sensor network change the way of human interacting with the physical world. Wireless sensor networks demonstrate potential applications in more and more areas due to its characteristics also will change people’s way of life; and the same time, more and more application of close to people’s live, the security issue has become increasingly people’s attention, and also gives researchers a lot of challenging issues.As the limited of the sensor nodes’ computing power and capacity of storage, traditional approaches to network security can not solve the problem about the security of WSN, further we can not completely copy the traditional approach to information security; it must find a new way suitable for the security of WSN. By way of bionics, conferring the node certain autonomy of human behavior is a good way to achieve sensor network security, and it is now a research hot spot. Introduced the concept of human credibility to the sensor networks, evaluate each node by trustworthiness of human society and measure the reliability of each node through it’s trusted value; if the node is not reliable will be isolate from the network in order to achieve sensor networks run secretly and smoothly.Based on the behavior of sensor nodes, the node’s reputation is divided into communication reputation, data reputation and evaluating reputation, three reputations have all adopted the Beta reputation system to carry on reputation management, and use the Josang’s belief model to integrate direct reputation and indirect reputation. In the study of other trust fusion algorithm, the paper presents a dynamic weight algorithm of trust fusion value, simulation results show that the new algorithm can well identify the malicious nodes in a single attack scenario, and has better recognition to multiple attacks. The paper also studies the forgotten mechanism and then proposes time window forgetting mechanism suitable for characteristics of the sensor network. In addition, the paper also analyzes the Beta reputation system; a rule is proposed that the formula for calculating the trust value of reputation system should meet feature of slowly ascend and quickly descend. On this basis, the dynamic adjustment trust value algorithm is proposed, and simulation results show that the new algorithm improves the ability of the reputation system detecting the attack. Correspondingly, the last punishment redemption mechanism in paper, which is related with the node’s continuous number of attacks, improves the reputation system’s sensitivity to attack.