Abstract
Key management has long been a challenging problem in wireless Distributed Sensor Networks (DSNs) due to their high security requirements and strict resource constraints. Recently, a randomized key pre-distribution scheme has been introduced to serve as a practical solution and many improvements are subsequently proposed. These schemes mainly focus on key allocations based on pre-deployment estimates of post-deployment information items, such as location data and attack probabilities. Unfortunately, such information items may be unavailable or may change over time. Based on adaptability to post-deployment contexts, we propose a general key redistribution framework that exploits neighboring keys from connected neighbors to reach unconnected nodes. We show that our framework can be applied to most existing key pre-distribution schemes (both key-based and polynomial-based) to further improve their performance. We demonstrate our proposed framework's salient features, such as high connectivity, high resilience, and efficient memory usage, by both analytical and simulation results.