Detection of time-varying directional flows in wireless networks

The problem of detecting packet flows between two nodes in a wireless network is considered. Especially, the transmission timings of two nodes are recorded, and their transmission rates can be time-varying (piecewise constant). Based on the timing measurements, our objective is to detect the presence of packet flows between them. Two different scenarios are considered; the first is that a flow may exist in only one specific direction, and the other is that a flow may exist in any direction. For each case, a detection algorithm is provided, and for the latter scenario, an additional algorithm aimed at estimating the direction of the underlying flow is proposed. When the transmission processes are nonhomogeneous Poisson processes, under certain conditions, our algorithms are proved to be consistent. The algorithms are tested using the MSN Voice over IP (VoIP) traffic and the synthetic Poisson traffic.

[1]  Vern Paxson,et al.  Multiscale Stepping-Stone Detection: Detecting Pairs of Jittered Interactive Streams by Exploiting Maximum Tolerable Delay , 2002, RAID.

[2]  Dawn Xiaodong Song,et al.  Detection of Interactive Stepping Stones: Algorithms and Confidence Bounds , 2004, RAID.

[3]  Stuart Staniford-Chen,et al.  Holding intruders accountable on the Internet , 1995, Proceedings 1995 IEEE Symposium on Security and Privacy.

[4]  Lang Tong,et al.  Timing-based detection of packet forwarding in MANETs , 2010, 2010 IEEE 11th International Workshop on Signal Processing Advances in Wireless Communications (SPAWC).

[5]  Lang Tong,et al.  Detection of Information Flows , 2008, IEEE Transactions on Information Theory.

[6]  Lang Tong,et al.  Detection of time-varying flows in wireless networks , 2010, 2010 - MILCOM 2010 MILITARY COMMUNICATIONS CONFERENCE.