Trust Score-Based Modeling – Experimental Code Repository

This repository provides the experimental code used in the paper: "A Trust Score-Based Access Control Model for Zero Trust Architecture: Design, Sensitivity Analysis, and Real-World Performance Evaluation"

---

Experiments

The repository contains three main types of experiments:

1. Cross-Dataset Performance Evaluation (UNSW-NB15 vs. CICIDS2017)

   • Evaluate the robustness of the TS model across different datasets.
   • Datasets: UNSW-NB15 (general-purpose attacks) and CICIDS2017 Wednesday subset (benign + DoS/DDoS).
   • Metrics: Accuracy, Precision, Recall, F1-Score, and ROC-AUC.

2. Computational Burden and Scalability of the TS Model

   • Measure latency, throughput, and memory usage.
   • Tested across dataset sizes up to 1,000,000 records to demonstrate scalability.

3. Comparative Benchmarking of Legacy Access Control Models vs. the TS Model

   • Models included:
     • RBAC (Role-Based Access Control)
     • ABAC (Attribute-Based Access Control)
     • RBA (Risk-Based Access Control)
   • Compared with the proposed Trust Score (TS) model.
   • Metrics: average latency (ms/record) and throughput (TPS).

---

Repository Structure

• evaluate_datasets_unsw_cicids.py
  Code for cross-dataset performance evaluation using UNSW-NB15 and CICIDS2017.

  • Includes preprocessing, train/test split, and classification performance analysis.

• ts_computational_burden.py
  Code for measuring computational burden and scalability of the TS model.

• access_control_models_unsw_v12.py
  Code for comparative benchmarking of RBAC, ABAC, RBA, and the TS model.

---

Dataset

• UNSW-NB15 dataset** (already included in this repository for reproducibility)
  Original reference: UNSW-NB15 – UNSW Canberra

• CICIDS2017 dataset** (Wednesday subset used for evaluation)
  Original reference: CICIDS2017 – Canadian Institute for Cybersecurity

※ The original datasets (UNSW-NB15 and CICIDS2017) are not included in this repository due to GitHub upload size limitations, but they can be obtained directly from the authors upon request.

---

Requirements

• Python 3.10+
• pandas
• numpy
• scikit-learn

Install required packages:

pip install pandas numpy scikit-learn