Research On Secure Key Management Schemes For Wireless Sensor Networks

Wireless sensor network(WSN) is highly vulnerable to attacks because it consists of numerous resource-constrained devices and communicates via wireless links. These vulnerabilities are more severe when WSN is deployed in an unattended hostile environment, such as battlefields. In such an environment, an adversary may physically capture and destroy any sensor nodes, then compromise critical security data including keys used for encryption and authentication. Consequently, it is very important to provide security services to WSN to ensure its survival. Researches in this paper focus on key management schemes for WSN.Firstly, an improved triple-key management scheme is presented. This scheme adopts the one-layer cluster model of LEACH protocol instead of the multi-layer cluster architecture in the original triple-key scheme, so that it simplifies the network topology, and saves some extra overheads caused by communication among multi-layer cluster headers, but it will inevitably exacerbate the communication frequency. Meanwhile, this scheme adjusts the secure routing algorithms to adapt it to the new network architecture and make the operation and implementation easier. Secondly, a novel key management scheme based on polynomial and grouping is presented, which has a nice tradeoff between performance indicators like network security, scalability, connectivity and overhead costs. This scheme uses shared keys by polynomial computing to enhance node's anti-attacking ability, and it introduces grouping idea to increase network connectivity. The unique grouping pattern divides node ID into two parts: group ID and intragroup node ID, thus we can enlarge the maximum supportable network size by extending the intragroup node ID.Experiments of the two schemes above are based on simulator OMNeT++ and language C++. Results show that the improved triple-key management sheme gets better network security and supports larger network scale, but it cannot allow adding sensor nodes dynamically. As for the latter scheme, the theoretical analysis and experimental researches show that this novel scheme gains much better security performance and is more practical, although it increases some network overheads.