|Stochastic Systems Group|
Distributed Flow Detection via Timing Analysis
School of Electrical and Computer Engineering, Cornell University
We consider the problem of detecting the presence of a packet flow in a wireless network using distributed traffic sensors that measure only the timing of packet transmissions. We first present a characterization of flow detectability. Specifically, under certain traffic conditions, there exists a sharp threshold on the flow rate above which the detection error probabilities diminish as the number packets used in the detection increases. On the other hand, the flow is undetectable if its rate is below the threshold. This threshold also characterizes the maximum flow rate that can be hidden in independent traffic realizations. We then present some results on the distributed detection when communications between traffic sensors and the fusion center are subject to rate constraints.
Joint work with Ting He at IBM T. J. Watson Research Center and Ameya Agaskar at MIT Lincoln Laboratory.
Problems with this site should be emailed to email@example.com