Deep Learning based Symbol Detection for Molecular Communications

Abstract

Molecular communication (MC) can play an indispensable role in nanonetworks and Internet of Bio-nano Things based applications. However, inter-symbol interference (ISI), due to slow diffusion of molecules can severely degrade system’s performance. In this paper, we propose a deep learning (DL)-based receiver design to decode the data symbols in MC. The proposed DL-based receiver (DLR) does not require the channel state information and threshold value(s) implicitly to decode the data symbols. The DLR is trained offline by applying the data symbols generated from simulation based on diffusion channel statistics, then it is used for recovering the online transmitted data symbols directly. Impact of various system parameters such as diffusion coefficient, noise and ISI level, and frame duration are analyzed for DLR. DLR’s performance is also compared to conventional detection methods. Results show that DLR can be a viable and practical choice in MC system design.