| In this thesis, we study performance and anonymity issues in wireless ad hoc networks, from the viewpoint of transmission power control. Based on the results of these studies, novel algorithms and protocols are proposed and evaluated in order to optimize network performance, with and without strict anonymity requirements. Our specific contributions are outlined as follows. (1) In Chapter 2, we present the analysis of eavesdropping risk, which is a measure of how much transmission power control affects the eavesdropping capability of adversaries in wireless ad hoc networks. We derive a closed-form formula for the first order eavesdropping risk as a function of the normalized transmission radius, when nodes are distributed uniformly over the network. When node distribution is not uniform, we derive the first known bound. The results of this analysis can be used as an effective means to weaken the eavesdropping capability of adversaries in wireless ad hoc networks, thus helping to defend against traffic analysis for which the greater the number of messages observed, the more can be inferred. (2) Using the results of the analysis of eavesdropping risk as an additional, complementary defense of line, the first anonymous unicast routing protocol which supports multiple transmission power/range is presented in Chapter 3. Compared with other anonymous unicast routing protocols, our proposed protocol reduces the end-to-end delay by orders of magnitude, while performing comparably well in terms of packet delivery ratio. The decrease in end-to-end delay is very important for many applications, especially for real-time traffic ( e.g., video and audio streaming) and interactive traffic. (3) In Chapter 4, energy-efficient multicast schemes designed for wireless ad hoc networks are analyzed. In practice, an energy-efficient multicast algorithm working at the network layer is often unable to be implemented due to the lack of router support; however, a multicast algorithm at the application layer, though easy to be implemented, typically causes severe performance degradation. The performance degradation stems from the fact that without router support, the information critical to energy efficiency most likely becomes unavailable. To solve this dilemma, we first formulate the application-layer energy-efficient multicast problem as the predictive energy-efficient overlay multicast problem. Then the missing information critical to energy efficiency is predicted, without knowing the actual value, by using our derived bounds for its corresponding mathematical expectation. Based on the predicted values rather than the actual values, we propose several low-complexity multicast algorithms and evaluate their energy efficiencies. Simulation results demonstrate significant energy savings, compared to existing energy-efficient multicast algorithms. (4) Finally, Chapter 5 expands the work presented in Chapter 4 to allow multicasting packets energy-efficiently and anonymously. In the past, strict anonymity requirements and energy-efficient multicast seemed contradictory goals because strict anonymity requirements prevented any energy-efficient multicast algorithm from knowing the information critical to energy efficiency and no prediction of such information was able to be made. Indeed, Chapter 5 presents and evaluates the first energy-efficient multicast algorithm that preserves strict anonymity requirements. Compared to the cases which solely run an anonymous unicast routing protocol, simulation results show a significant performance improvement in terms of energy efficiency, throughput, delay, etc. |
