They dont care about which node makes the false modification or injection But

They dont care about which node makes the false

This preview shows page 6 - 8 out of 10 pages.

They don’t care about which node makes the false modification or injection. But the information created in the filtering process can be used to locate the misbehaving node. For example, [24] proposed that a couple of nodes did the same fusing as the witness of each other. The base station decided whether a fusing result was legitimate by an m out of n scheme. Cryptography was used to prevent intermediate modification and assure the node’s identification. Although the malfunction node could be detected by this way, no further punishment would be carried out to the wrongdoer. The way to deal with false data or injected data was simply discarding. In the behavior-based trust model, the trust degree of the wrongdoer will be degraded, and its following report is less believable. 5 Combination Framework The behavior evaluation methods are task-specific, and differ from each other. To a given task, the evaluation result may be accurate or inaccurate. And sometimes there is no accessible evaluation at all. So it’s very difficult to combine these measurements into an integrated trust degree that reflects the essential trustworthiness of the node. A naïve combination method is to record the result of recent tasks, i.e., the number of good and bad behavior, the number of total tasks. If the ratio of good behavior to total numbers surpasses a certain threshold, the node is thought to be trustworthy and is qualified for future assignment. Otherwise, the node will be excluded from future tasks as long as there are alternatives. This simple method has several drawbacks: 1) It treats
Image of page 6
220 L. Huang, L. Li, and Q. Tan all the measurement as a two-value function, which is not sufficient under some situation; 2) All evaluation types are treated equally, while in practice some are more important than others; 3) This method can’t deal effectively with the second-hand information such as the trust degree given by other node. Combination algorithms from “Mathematic Theory of Evidence [25]” can serve the purpose better, and actually they have been used in some researches. 5.1 Bayesian Inference Bayesian inference is used in [26] to combine the observations. The behavior was described by a binary variable {cooperative, uncooperative} . The reputation, R ij =P(node j is trustworthy at node j) ,was calculated from past observations. It assumed that R ij obeyed the beta distribution, R ij =Beta( α i , β i ) , and initial α 0 = β 0 =1. After α i cooperative and β i uncooperative behaviors had been observed, the reputation was refreshed to be R ij =Beta( α i +1, β i +1) . The impact of second-hand information was also expressed by a new α and β which took the trust value of the source into consideration. The trust value of a node was T=E[R ij ] . 5.2 Dempster-Shafer Theory of Evidence The DS theory of evidence was first introduced by Dempster and formalized by Shafer. Suppose there is a focal set of mutually exclusive and exhaustive propositions, which is also called frame of discernment F . The inference space Θ is a power set of F . For F={a,b,c} , Θ ={a.b.c,{a,b},{a,c},{b,c},{a,b,c},Ø} . Mass function
Image of page 7
Image of page 8

You've reached the end of your free preview.

Want to read all 10 pages?

  • Fall '19
  • Sensor node, Wireless sensor network

  • Left Quote Icon

    Student Picture

  • Left Quote Icon

    Student Picture

  • Left Quote Icon

    Student Picture