Signal Detection of Cooperative Multi-Hop Mobile Molecular Communication via Diffusion

Abstract

The data-driven detectors based on deep learning have promising applications in signal detection with unknown channel parameters of molecular communication via diffusion (MCvD) system. In this paper, a signal detector for cooperative multi-hop mobile MCvD system with amplify-forward relaying strategy by using Transformer-based model is proposed. The mathematical expressions of the numbers of received molecules when considering two transmission schemes including multi-molecule-type (MMT) and single-molecule-type (SMT) are derived in order to generate the training dataset. On this basis, the training dataset is used to train the Transformer-based model offline. Then the trained Transformer-based model is adopted to detect the received signal under unknown channel parameters under MMT and SMT. Numerical results show that the Transformer-based model performs the best detection ability in cooperative multi-hop mobile MCvD system with lowest bit error rate of signal detection compared with deep neural networks (DNN) detector and convolutional neural networks (CNN) detector.