| A common deployment strategy for wireless sensor networks (WSN) is to position sensors according to a probabilistic distribution. Some sensor distributions offer advantages in redundancy, which helps ensure network survivability even with the loss of multiple sensor nodes to cyber attack. Properly selecting an initial sensor deployment mitigates the threat of a denial of service (DoS) attack. While it is clear that some distributions are superior to others, it is not clear how to select the best sensor distribution as a defensive measure. Current strategies rely on mathematical analysis and are restricted to a subset of possible distributions. This thesis examines the problem from an experimental perspective. It uses a novel, agent-based model to simulate WSN under DoS attack. This model informs a decision support system for selecting a resilient, energy-efficient sensor distribution. Results show that routing protocol and sensor placement have substantial impact on the behavior and lifetime of WSN during cyber attack; analyzing this behavior aids in deciding on a sensor deployment. |
