Wireless Sensor Networks (WSNs) have been utilized for many applications such as tracking and monitoring of endangered species in a national park, soldiers in a battlefield, and many others, which require anonymity of the origin, known as the Source Location Privacy (SLP ). The aim of SLP is to prevent unauthorized observers from tracing the source of a real event (an asset) by analyzing the traffic of the network. We develop the following six techniques to provide anonymity: Dummy Uniform Distribution (DUD), Dummy Adaptive Distribution (DAD), Controlled Dummy Adaptive Distribution (CAD), Exponential Dummy Adaptive Distribution (EDAD), Exponential Dummy Adaptive Distribution Plus One (EDADP1), and Exponential Dummy Adaptive Distribution Plus Two (EDADP2). Moreover, an enhanced version of the well-known FitProbRate technique is also developed. The purpose of these techniques is to overcome the anonymity problem against a global adversary model that has the capability of analyzing and monitoring the entire network.;We perform an extensive verification of the proposed techniques via simulation, statistical, and visualization approaches. Three analytical models are developed to verify the performance of our techniques: A Visualization model is performed on the simulation data to confirm anonymity. A Neural Network model is developed to ensure that the introduced techniques preserve SLP. In addition, a Steganography model based on statistical empirical data is implemented to validate the anonymity of the proposed techniques. The Simulation demonstrates that the proposed techniques provide a reasonable delay, delivery ratio, and overhead of the real event's packets while keeping a high level of anonymity.;Results show that the improved version of FitProbRate massively reduces the number of operations needed to detect the distribution type of a data sequence despite the number of intervals when compared to the original. A comprehensive comparison between EDADP1, EDADP2, and FitProbRate in terms of the average delay, anonymity level, average processing time, Anderson-Darling test, and polluted scenarios is conducted. Results show that all three techniques have a similar performance regarding the average delay and Anderson-Darling test. However, the proposed techniques outperform FitProbRate in terms of anonymity level, average processing time, and polluted scenarios. WSN applications that need privacy can select the suitable proposed technique based on the required level of anonymity with respect to delay, delivery ratio, and overhead.
展开▼
