Free Trial

Safari Books Online is a digital library providing on-demand subscription access to thousands of learning resources.

Share this Page URL

8.6.6 Localizing Adversaries > Experimental Evaluation - Pg. 217

CHAPTER 8 Defending Against Identity-Based Attacks in Wireless Networks 4. The Solver instance performs localization and returns the coordinates of the transmitter back to the Server. If there is a need to localize hundreds of trans- mitters at the same time, the server can perform load balancing among different solver instances. This centralized localization solution also makes enforcing contracts and privacy policies more tractable. Attack Localizer. When the detector has identi- fied an attack for a node identity, the centroids returned by the K-means clustering analysis in signal space can be used by the server and sent to the solver for location estimation. Particularly, in spoofing attacks the returned positions should be the location estimate for the original node and the spoofing nodes in physical space. In order to show the generality of the proposed system for locating the adversarial nodes, we 217 of area-based methods compared to point-based methods is that they return a region, which has an increased chance of capturing the transmit- ter's true location. ABP returns an area, a set of tiles on the floor, bounded by a probability that the transmitter is within the returned area [43]. ABP assumes the distribution of RSS for each landmark follows a Gaussian distribution. The Gaussian random variable from each landmark is independent. ABP then computes the probabil- ity of the transmitter being at each tile L i on the floor using Bayes' rule: Equation 8-34 P (L i |s) = P (s|L i ) × P (L i ) . P (s) Given that the transmitter must reside at exactly one tile satisfying L =1 P (L i |s) = 1, ABP i normalizes the probability and returns the most