owards a Statistical Framework for Source Anonymity in Sensor Networks.
To improve anonymity, we suggest introducing the same correlation of inter-transmission times during real intervals to inter-transmission times during fake intervals. That is, let the transmission procedure consists of two different algorithms: AR and AF . In the presence of real events , algorithm AR is implemented. In the absence of real events , algorithm AF is implemented. In algorithm AF , the nodes generates two sets of events independently of each other: “dummy events” and fake events. Fake events serve the same purpose they serve in algorithm AR, that is, they are used to hide the existence of real transmissions. Since there are no real events in fake intervals, however, dummy events are generated to be handled as if they are real events. That is, dummy events are generated independently of fake messages and, upon their generation, their transmission times are determined according to the used statistical goodness of fit test. The purpose of this procedure is to introduce the same correlation of real intervals into fake intervals. That is, not only the two sequences of intertransmission times will be statistically indistinguishable by means of statistical goodness of fit tests, but also the binary codes representing fake and real intervals will have the same statistical behavior.