Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Clustering Guided Residual Neural Networks for Multi-Tx Localization in Molecular Communications

Ali Sonmez, Erencem Ozbey, Efe Feyzi Mantaroglu, H. Birkan Yilmaz

TL;DR

Localization of multiple transmitters in MCvD is challenging due to Brownian diffusion and overlapping molecule distributions on the receiver surface. The authors propose clustering-based centroid corrections to standard K-means and two clustering-guided Residual Neural Networks, AngleNN for direction and SizeNN for cluster size estimation, to refine initial estimates. Key contributions include MinCovDet, Density-Weighted Centroid, Density-MinCovDet corrections, and AngleNN/SizeNN architectures with a 4K input, 256-dim stem, six residual blocks. Experiments show substantial reductions in localization error relative to K-means across 2-, 3-, and 4-Tx scenarios (e.g., up to 73% angular error reduction with AngleNN and 69% MAPe reduction with AngleNN+SizeNN), along with improved size estimation. This work enhances robustness to density variations and outliers, enabling more reliable multi-Tx localization in MCvD.

Abstract

Transmitter localization in Molecular Communication via Diffusion is a critical topic with many applications. However, accurate localization of multiple transmitters is a challenging problem due to the stochastic nature of diffusion and overlapping molecule distributions at the receiver surface. To address these issues, we introduce clustering-based centroid correction methods that enhance robustness against density variations, and outliers. In addition, we propose two clusteringguided Residual Neural Networks, namely AngleNN for direction refinement and SizeNN for cluster size estimation. Experimental results show that both approaches provide significant improvements with reducing localization error between 69% (2-Tx) and 43% (4-Tx) compared to the K-means.

Clustering Guided Residual Neural Networks for Multi-Tx Localization in Molecular Communications

TL;DR

Localization of multiple transmitters in MCvD is challenging due to Brownian diffusion and overlapping molecule distributions on the receiver surface. The authors propose clustering-based centroid corrections to standard K-means and two clustering-guided Residual Neural Networks, AngleNN for direction and SizeNN for cluster size estimation, to refine initial estimates. Key contributions include MinCovDet, Density-Weighted Centroid, Density-MinCovDet corrections, and AngleNN/SizeNN architectures with a 4K input, 256-dim stem, six residual blocks. Experiments show substantial reductions in localization error relative to K-means across 2-, 3-, and 4-Tx scenarios (e.g., up to 73% angular error reduction with AngleNN and 69% MAPe reduction with AngleNN+SizeNN), along with improved size estimation. This work enhances robustness to density variations and outliers, enabling more reliable multi-Tx localization in MCvD.

Abstract

Transmitter localization in Molecular Communication via Diffusion is a critical topic with many applications. However, accurate localization of multiple transmitters is a challenging problem due to the stochastic nature of diffusion and overlapping molecule distributions at the receiver surface. To address these issues, we introduce clustering-based centroid correction methods that enhance robustness against density variations, and outliers. In addition, we propose two clusteringguided Residual Neural Networks, namely AngleNN for direction refinement and SizeNN for cluster size estimation. Experimental results show that both approaches provide significant improvements with reducing localization error between 69% (2-Tx) and 43% (4-Tx) compared to the K-means.

Paper Structure

This paper contains 7 sections, 5 equations, 4 figures, 3 tables.

Table of Contents

  1. Introduction
  2. System Model
  3. Methodology
  4. Algorithmic Correction of Cluster Centers
  5. Refinement with Residual Networks
  6. Performance Evaluation
  7. Conclusion

Figures (4)

  • Figure 1: A Multi-Tx system where received molecules are illustrated. Emitted molecules from different point-source Tx's create overlapping distributions on the spherical Rx, making the separation and localization more challenging.
  • Figure 2: Mean angular error (in degrees) for different methods under 2-Tx, 3-Tx and 4-Tx scenarios.
  • Figure 3: Empirical cumulative distribution functions (ECDFs) of angular errors in 2-Tx, 3-Tx, and 4-Tx scenarios.
  • Figure 4: Localization performance under varying cluster size imbalance for 2-Tx.