Extending a Parallel Boundary-Based CPM Algorithm Using a GPU-Accelerated Reaction-Diffusion Solver for Molecular Communication Simulations

Molecular communication (MC) is an emerging paradigm in which bio-nanomachines exchange information via the diffusion of signaling molecules. To simulate MC systems, the Cellular Potts Model (CPM) provides a flexible framework that captures the morphological dynamics and spatial interactions of bio-nanomachines. In this paper, we extend the boundary-based (BL) algorithm of the CPM with a CPU–GPU hybrid framework to efficiently support MC simulations. Specifically, we incorporate a GPU–based reaction-diffusion partial differential equation (PDE) solver to model the propagation and degradation of signaling molecules emitted by transmitter bio-nanomachines. In the example simulations, receiver bio-nanomachines exhibit chemotactic migration toward the transmitter region, demonstrating that the proposed framework can simulate molecular communication in a drug delivery system (DDS) scenario. In addition, we show that the proposed framework reduces computational runtime compared with a CPU-based implementation.