A Fault Tolerant Parallel Computing Scheme of Scalar Multiplication for Wireless Sensor Networks

In event-driven sensor networks, when a critical event occurs, sensors should transmit messages back to base station in a secure and reliable manner. We choose Elliptic Curve Cryptography to secure the network since it offers faster computation and good security with shorter keys. In order to minimize the running time, we propose to split and distribute the computation of scalar multiplications by involving neighboring nodes in this operation. In order to improve the reliability, we have also proposed a fault tolerance mechanism. It uses half of the available cluster members as backup nodes which take over the work of faulty nodes in case of system failure. Parallel computing does consume more resources, but the results of simulation show that the computation can be significantly accelerated. This method is designed specially for applications where running time is the most important factor.