Key Predistribution Using Combinatorial Designs for Grid-Group Deployment Scheme in Wireless Sensor Networks

We propose a new grid-group deployment scheme in wireless sensor networks. We use combinatorial designs for key predistribution in sensor nodes. The deployment region is divided into square regions. The predistribution scheme has the advantage that all nodes within a particular region can communicate with each other directly and nodes which lie in a different regions can communicate via special nodes called agents which have more resources than the general nodes. The number of agents in a region is always three, whatever the size of the network. We give measures of resiliency taking the Lee distance into account. Apart from considering the resiliency in terms of fraction of links broken, we also consider the resiliency as the number of nodes and regions disconnected when some sensor are compromised. This second measure, though very important, had not been studied so far in key predistribution schemes which use deployment knowledge. We find that the resiliency as the fraction of links compromised is better than existing schemes. The number of keys preloaded in each sensor node is much less than all existing schemes and nodes are either directly connected or connected via two hop paths. The deterministic key predistribution schemes result in constant-time computation overhead for shared key discovery and path key establishment.