Simulation of Brownian Motion for Molecular Communications on a Graphics Processing Unit

This work applies a graphics processing unit (GPU) to the study of molecular communication (MC) systems where molecules are used to exchange information. Most MC is based on Brownian motion and modeled via a stochastic differential equation that admits analytical solutions under certain rather restrictive assumptions. As such, emphasis is placed on Monte Carlo simulation methods to study MC. This paper explores the application of a GPU to reduce this simulation time using two different approaches. This work will show that the GPU can offer significant speedup relative to the CPU, providing avenues to deeper MC research unavailable using a CPU based simulator. With that, it will also show that some avenues towards deeper MC research remain infeasible due to excessively long simulation times, even when using a GPU.