Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Machine Learning-Based Localization Accuracy of RFID Sensor Networks via RSSI Decision Trees and CAD Modeling for Defense Applications

Curtis Lee Shull, Merrick Green

TL;DR

This work investigates zone-level indoor localization of defense assets using RSSI data from RFID in a CAD-modeled lab. A Decision Tree classifier is trained on ReaderIP, Antenna, and RSSI to predict one of 12 zones, with class weighting used to mitigate imbalance. The model achieves 34.2% overall accuracy and 58.7% adjacency-aware accuracy, highlighting that high-coverage zones are more reliably inferred while rare zones remain challenging. The study demonstrates the value of ML-enabled simulations for defense logistics, identifies practical limitations, and points to sensor fusion and CAD-informed antenna planning as avenues to improve robustness for mission-critical asset tracking.

Abstract

Radio Frequency Identification (RFID) tracking may be a viable solution for defense assets that must be stored in accordance with security guidelines. However, poor sensor specificity (vulnerabilities include long range detection, spoofing, and counterfeiting) can lead to erroneous detection and operational security events. We present a supervised learning simulation with realistic Received Signal Strength Indicator (RSSI) data and Decision Tree classification in a Computer Assisted Design (CAD)-modeled floor plan that encapsulates some of the challenges encountered in defense storage. In this work, we focused on classifying 12 lab zones (LabZoneA-L) to perform location inference. The raw dataset had approximately 980,000 reads. Class frequencies were imbalanced, and class weights were calculated to account for class imbalance in this multi-class setting. The model, trained on stratified subsamples to 5,000 balanced observations, yielded an overall accuracy of 34.2% and F1-scores greater than 0.40 for multiple zones (Zones F, G, H, etc.). However, rare classes (most notably LabZoneC) were often misclassified, even with the use of class weights. An adjacency-aware confusion matrix was calculated to allow better interpretation of physically adjacent zones. These results suggest that RSSI-based decision trees can be applied in realistic simulations to enable zone-level anomaly detection or misplacement monitoring for defense supply logistics. Reliable classification performance in low-coverage and low-signal zones could be improved with better antenna placement or additional sensors and sensor fusion with other modalities.

Machine Learning-Based Localization Accuracy of RFID Sensor Networks via RSSI Decision Trees and CAD Modeling for Defense Applications

TL;DR

This work investigates zone-level indoor localization of defense assets using RSSI data from RFID in a CAD-modeled lab. A Decision Tree classifier is trained on ReaderIP, Antenna, and RSSI to predict one of 12 zones, with class weighting used to mitigate imbalance. The model achieves 34.2% overall accuracy and 58.7% adjacency-aware accuracy, highlighting that high-coverage zones are more reliably inferred while rare zones remain challenging. The study demonstrates the value of ML-enabled simulations for defense logistics, identifies practical limitations, and points to sensor fusion and CAD-informed antenna planning as avenues to improve robustness for mission-critical asset tracking.

Abstract

Radio Frequency Identification (RFID) tracking may be a viable solution for defense assets that must be stored in accordance with security guidelines. However, poor sensor specificity (vulnerabilities include long range detection, spoofing, and counterfeiting) can lead to erroneous detection and operational security events. We present a supervised learning simulation with realistic Received Signal Strength Indicator (RSSI) data and Decision Tree classification in a Computer Assisted Design (CAD)-modeled floor plan that encapsulates some of the challenges encountered in defense storage. In this work, we focused on classifying 12 lab zones (LabZoneA-L) to perform location inference. The raw dataset had approximately 980,000 reads. Class frequencies were imbalanced, and class weights were calculated to account for class imbalance in this multi-class setting. The model, trained on stratified subsamples to 5,000 balanced observations, yielded an overall accuracy of 34.2% and F1-scores greater than 0.40 for multiple zones (Zones F, G, H, etc.). However, rare classes (most notably LabZoneC) were often misclassified, even with the use of class weights. An adjacency-aware confusion matrix was calculated to allow better interpretation of physically adjacent zones. These results suggest that RSSI-based decision trees can be applied in realistic simulations to enable zone-level anomaly detection or misplacement monitoring for defense supply logistics. Reliable classification performance in low-coverage and low-signal zones could be improved with better antenna placement or additional sensors and sensor fusion with other modalities.
Paper Structure (33 sections, 13 equations, 5 figures, 1 table)

This paper contains 33 sections, 13 equations, 5 figures, 1 table.

Table of Contents

  1. Introduction
  2. Related Work
  3. Methodology
  4. Class-weight computation.
  5. Tag-to-zone mapping.
  6. Split protocol.
  7. Results and Evaluation
  8. Zone-Level Classification Performance
  9. Per-zone heterogeneity.
  10. Impact of class weights.
  11. Visual Summaries
  12. ReaderIP vs RSSI scatter (Fig. \ref{['fig:readerip']}).
  13. RSSI boxplots (Fig. \ref{['fig:boxplot']}).
  14. RSSI distributions (Fig. \ref{['fig:distribution']}).
  15. Confusion matrix (Fig. \ref{['fig:confusion']}).
  16. ...and 18 more sections

Figures (5)

  • Figure 1: Distribution of raw zone samples (bars, left axis) and the resulting computed class weights $w_k$ (line, right axis). Common zones (e.g. G, K, E) are assigned a weight of less than 1.0 to down-weight their influence while rare zones (A, D, C) are up-weighted. LabZoneC is given the largest adjustment ($w=1.80$).
  • Figure 2: Scatter plot of Reader IP versus RSSI showing clustered signal profiles by antenna.
  • Figure 3: Boxplot of RSSI values across all 12 zones.
  • Figure 4: RSSI distribution histogram for key zones.
  • Figure 5: Confusion matrix of predicted vs actual zones.