A novel traffic analysis attack model and base‐station anonymity metrics for wireless sensor networks

Abstract

In applications of wireless sensor networks (WSNs), all data packets are directed toward the base-station (BS) over multihop routes. In addition to data processing, the BS can interface the WSN to remote centers. Therefore, the failure of the BS diminishes the network utility and makes the data inaccessible. An adversary would thus try to uncover the BS's location by analyzing the traffic patterns of the network in order to launch denial of service attacks. For that reason, various countermeasure techniques have been proposed to boost the anonymity of the BS. In this paper, we study the existing traffic analysis models that an adversary could apply to locate the BS and point out shortcomings that make contemporary countermeasures weaker than thought. We further present a novel attack model that increases the adversary's confidence in localizing the BS while significantly reducing the traffic analysis complexity. Additionally, we introduce two novel anonymity metrics that can be used for evaluating the network resilience to attacks and for gauging the performance of countermeasures. The simulation results confirm the effectiveness of the proposed attack model.

Traffic Analysis Through Spatial and Temporal Correlation: Threat and Countermeasure

“…Assessing the BS anonymity has been the focus of studies such as [8][12]- [24] [63] and is in general a byproduct of the traffic analysis model. In this paper, ET is pursued as the underlying traffic analysis model where each intercepted transmission is deemed as an evidence of direct communication between a sender-receiver pair; the sender is determined by localizing the transmitter, while the receiver could be any node in a set of potential candidates within the sender's reachable range.…”

“…Thus, we believe that any modification of ET and its Belief measure based on the length of a derived path might introduce error and falsify the statistical significance of said path. Therefore, in this paper we do not use the weighted Belief measure that has been adopted in recent work [17]- [23][27] [63].…”

“…An adversary can easily filter out/in the cells of a grid based on their distinguishability (high value or low value) and implement targeted attack on them; e.g. Zoom strategy [16] or EARS [63]. Thus, a low value of STDEV is the sign of better anonymity.…”

Efficient yet Robust Privacy Preservation for MPEG-DASH-Based Video Streaming

“…Since most web traffic is encrypted these days, most traffic analysis attacks often rely on side channel information and machine learning to be effective. Some side channel reliant traffic analysis attacks use only timing information [20,27] while others depend on different side channel information, such as presence or absence of communication information [28], delaying and analyzing HTTP requests [27], and standard side channel information, such as packet length, number of packets, and time [25]. Machine learning allows adversaries to analyze even encrypted traffic to steal user information [15,[29][30][31].…”

Secure Mobile Sink Node location in Wireless Sensor Network using Dynamic Routing Protocol

“…On the other direction, Baroutis and Younis showed in their study [5] how the attackers perform traffic analysis using three kinds of attack models: GSAT Test, Entropy and Traffic Volume (TV) and Evidence Theory (ET). In the GSAT Test, the attackers follow a number of steps until they discover the SN location.…”