A Hybrid Key Predistribution Scheme for Sensor Networks Employing Spatial Retreats to Cope with Jamming Attacks

In order to provide security services in wireless sensor networks, a well-known task is to provide cryptographic keys to sensor nodes prior to deployment. It is difficult to assign secret keys for all pairs of sensor node when the number of nodes is large due to the large numbers of keys required and limited memory resources of sensor nodes. One possible solution is to randomly assign a few keys to sensor nodes and have nodes be able to connect to each other with some probability. This scheme has limitations in terms of the tradeoffs between connectivity and memory requirements. Recently, sensor deployment knowledge has been used to improve the level of connectivity while using lesser amounts of memory space. However, deployment based key predistribution schemes may cause a large number of nodes to be cryptographically isolated if nodes move after key pre-distribution. Mobility may be necessitated for reasons depending on applications or scenarios. In this paper, we consider mobility due to spatial retreat of nodes under jamming attacks as an example. Jamming attacks are easy and efficient means for disruption of the connectivity of sensors and thus the operation of a sensor network. One solution for mobile sensor nodes to overcome the impact of jamming is to perform spatial retreats by moving nodes away from jammed regions. Moved nodes may not be able to reconnect to the network because they do not have any shared secret with new neighbors at new locations if strict deployment knowledge based key predistribution is employed. In this paper, we propose a hybrid key predistribution scheme that supports spatial retreat strategies to cope with jamming attacks. Our scheme combines the properties of random and deployment knowledge based key predistribution schemes. In the presence of jamming attacks, our scheme provides high key connectivity (similar to deployment knowledge based schemes) while reducing the number of isolated nodes. We evaluate the performance of our scheme through simulations and analysis.