GPU-based parallel computing of energy consumption in wireless sensor networks

The lifetime of a wireless sensor network is the most important design parameter to take into account. Given the autonomous nature of the sensor nodes, this period is mainly related to their energy consumption. Hence, the high interest to evaluate through accurate and rapid simulations the energy consumption for this kind of networks. However, in the case of a network with several thousand nodes, the simulation time can be very slow and even impossible in some cases. In this paper, we present a new model for a parallel computing of energy consumption in wireless sensor networks. This model is combined with a discrete event simulation in a multi-agent environment and implemented on GPU architecture. The results show that the proposed model provides simulation times significantly faster than those obtained by the sequential model for large networks and for long simulations. This improvement is more significant if the processing on each node is very time consuming. Finally, the proposed model has been fully integrated and validated into the CupCarbon simulator.