Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Autoencoder-based Optimization of Multi-user Molecule Mixture Communication Systems

Bastian Heinlein, Nuria Zurita Jiménez, Kaikai Zhu, Sümeyye Carkit-Yilmaz, Robert Schober, Vahid Jamali, Maximilian Schäfer

Abstract

In this paper, we introduce an autoencoder (AE)-based scheme for end-to-end optimization of a multi-user molecule mixture communication system. In the proposed scheme, each transmitter leverages an encoder network that maps the user symbol to a molecule mixture. The mixtures then propagate through the channel to the receiver, which samples the channel using a non-linear, cross-reactive sensor array. A decoder network then estimates the symbol transmitted by each user based on the sensor observations. The proposed scheme achieves, for a given signal-to-noise ratio, lower symbol error rates than a baseline scheme from the literature in a single-user setting with full channel state information. We additionally demonstrate that the proposed AE-based scheme allows reliable communication when the channel is unknown or changing. Finally, we show that for multiple access the system can account for different user priorities. In summary, the proposed AE-based scheme enables end-to-end system optimization in complex scenarios unsuitable for analytical treatment and thereby brings molecular communication systems closer to real-world deployment.

Autoencoder-based Optimization of Multi-user Molecule Mixture Communication Systems

Abstract

In this paper, we introduce an autoencoder (AE)-based scheme for end-to-end optimization of a multi-user molecule mixture communication system. In the proposed scheme, each transmitter leverages an encoder network that maps the user symbol to a molecule mixture. The mixtures then propagate through the channel to the receiver, which samples the channel using a non-linear, cross-reactive sensor array. A decoder network then estimates the symbol transmitted by each user based on the sensor observations. The proposed scheme achieves, for a given signal-to-noise ratio, lower symbol error rates than a baseline scheme from the literature in a single-user setting with full channel state information. We additionally demonstrate that the proposed AE-based scheme allows reliable communication when the channel is unknown or changing. Finally, we show that for multiple access the system can account for different user priorities. In summary, the proposed AE-based scheme enables end-to-end system optimization in complex scenarios unsuitable for analytical treatment and thereby brings molecular communication systems closer to real-world deployment.
Paper Structure (11 sections, 2 equations, 5 figures)

This paper contains 11 sections, 2 equations, 5 figures.

Table of Contents

  1. Introduction
  2. System Model
  3. Auto-Encoder-based End-to-End Optimization
  4. Auto-Encoder Architecture
  5. Loss Function
  6. Evaluation
  7. Simulation Setup
  8. Static Point-to-Point Link with Full CSI
  9. Unknown Channel Attenuation
  10. Multi-User Scenario
  11. Conclusion and Future Work

Figures (5)

  • Figure 1: System Overview. We consider a scenario where multiple TXs transmit information to a single RX. At the $i$-th TX, the corresponding encoder determines which mixture $\mathbf{\bar{x}}_i[k]$ should be released for symbol $s[k]$ and sends the corresponding control signal to the ODD (ODD). Due to hardware imperfections, the ODDs release mixtures $\mathbf{x}_i[k]$ which propagate through the channel, resulting in a molecule mixture $\mathbf{y}[k]$ at the RX. The decoder then leverages the output of the RX sensor array $\mathbf{z}[k]$ to compute symbol estimates $\hat{s}_1[k], \dots, \hat{s}_U[k]$.
  • Figure 2: Full CSI. We show the SER of the proposed AE-based scheme (solid) and the baseline scheme from heinlein:nanocom (dotted) as a function of $1/\nu$ for different alphabet sizes $N$ (indicated by different colors).
  • Figure 3: Sensor Outputs for Unknown Channel Attenuation. We show the sensor outputs (dots) for the different symbols (indicated by different colors). The solid, dashed, and dotted lines indicate respectively the areas where $95\%$ of the symbols for $h=0.02$, $h \in \mathcal{H}_{\mathrm{lim}}$, and $h \in \mathcal{H}_{\mathrm{full}}$ lie.
  • Figure 4: SER for different channel attenuations. We show the SER of four different schemes as a function of the channel attenuation $h$ and different alphabet sizes $N$.
  • Figure 5: User Importance. We compare the user-specific SER and the SSER (SSER) for AE-based end-to-end optimization and CSK modulation of each user for various importance factors.