Neural Network Detectors for Molecular Communication Systems

Abstract

We consider molecular communication systems and show it is possible to train detectors without any knowledge of the underlying channel models. In particular, we demonstrate that a technique we previously developed, which is called sliding bidirectional recurrent neural network (SBRNN), performs well for a wide range of channel states when it is trained using a dataset that contains many sample transmissions under various channel conditions. We also demonstrate that the bit error rate (BER) performance of the proposed SBRNN detector is better than that of a Viterbi detector (VD) with imperfect channel state information (CSI) and it is computationally efficient.