Data for Near optimal routing protocols for source location privacy in wireless sensor networks : modelling, design and evaluation

Wireless Sensor Networks (WSNs) are collections of small computing devices that are used to monitor valuable assets such as endangered animals. As WSNs communicate wirelessly they leak information to malicious eavesdroppers. When monitoring assets it is important to provide Source Location Privacy (SLP), where the location of the message source must be kept hidden. Many SLP protocols have been developed by designing a protocol using intuition before evaluating its performance. However, this does not provide insight into how to develop optimal approaches. This thesis will present an alternate approach where the SLP problem is modelled using different techniques to give an optimal output. However, as this optimal output is typically for a restricted scenario, algorithms that trade optimality for generality are subsequently designed. Four main contributions are presented. First, an analysis is performed based on entropy and divergence to gain insight into how to reduce the information an attacker gains via the use of competing paths, and ways to compare the information loss of arbitrary routing protocols. Secondly, the SLP problem is modelled using Integer Linear Programming. The model result guides the design of a generic protocol called ILPRouting that groups messages together to reduce the moves an attacker makes. Thirdly, a timing analysis of when events occur is used to dynamically determine fake source parameters for the Dynamic and DynamicSPR algorithms. These fake sources lure the attacker to their location instead of the real source. Finally, the first SLP-aware duty cycle is investigated, and implemented for DynamicSPR to make it more energy efficient. These techniques are evaluated through simulations and deployments on WSN testbeds to demonstrate their effectiveness