Adaptive key management for wireless sensor networks

Network security protocols for wireless sensor networks (WSN) require each node to store context information, such as cryptographic keys, for each active security association that it maintains with its peers. While the storage requirements for such context information is generally small enough to be overlooked in regular networks, it becomes an important issue in memory-constrained WSN. Under such constraints, a WSN node may become memory saturated and thereby become unable to accommodate new nodes dynamically. What can be done to keep expanding the network with newly joining nodes under such memory constraints? How many security associations are `enough´ and how does the choice of the number of associations affect performance and coverage? To address these issues, we introduce a novel authentication and key management mechanism called Adaptive Key Management (AKM) that is robust and scalable under limited memory constraints and works along with the RPL routing protocol to provide strong security guarantees. We present a simulation study of AKM to demonstrate its scalability and to provide insight on how to allocate a node´s memory to store a sufficient number of security associations.