SDN-Based Multipath Data Offloading Using Link Quality Prediction

This repository implements the LSTM-based channel quality prediction methodology described in the paper: SDN-Based Multipath Data Offloading Scheme Using Link Quality Prediction for LTE and WiFi Networks. The implementation focuses on predicting channel quality using machine learning models (LSTM and BLSTM) to aid in decision-making for data offloading in heterogeneous LTE-WiFi networks.

---

Key Highlights

• Link Quality Prediction: Uses LSTM and BLSTM models for classifying channel quality into Good, Intermediate, and Bad categories.
• Metrics Used:
  • Hardware-based: Received Signal Strength Indicator (RSSI).
  • Software-based: Packet Delivery Ratio (PDR).
• Application Context: This prediction is intended to assist SDN-based traffic offloading decisions between LTE and WiFi.

---

Motivation

The rapid growth of mobile data traffic demands efficient utilization of network resources. This project:

• Implements link quality prediction using deep learning to classify and forecast channel conditions.
• Enables smarter decision-making for multipath data offloading schemes in SDN-enabled LTE-WiFi networks.

---

Methodology

1. Link Quality Prediction

• Models Implemented:
  • Long Short-Term Memory (LSTM).
  • Bidirectional LSTM (BLSTM).
• Input Metrics:
  • RSSI (hardware-based).
  • PDR (software-based).
• Classes:
  • Good, Intermediate, Bad.

2. Data Preprocessing

• Dataset is derived from IoT-LAB, with features such as RSSI and PDR.
• Preprocessing includes:
  • Removing outliers and redundancies.
  • Standardizing features using StandardScaler from scikit-learn.

3. Training and Prediction

• LSTM and BLSTM models are trained on an 80-20 train-test split.
• Prediction performance is evaluated using metrics like accuracy and confusion matrices.

---

Results

• Prediction Accuracy:
  • LSTM: 99.73%.
  • BLSTM: 99.94%.
  • Combined use of RSSI and PDR outperforms individual metrics.

---

Requirements

• Python 3.9+
• Libraries: tensorflow, scikit-learn, matplotlib

Citation 📝

If you use the code in this repository, please cite the following paper:

@article{kamath2024sdn,
  title={SDN-Based Multipath Data Offloading Scheme Using Link Quality Prediction for LTE and WiFi Networks},
  author={Kamath, Santhosha and Raman, J. Aravinda and Kumar, Pankaj and Singh, Sanjay and Kumar, M. Sathish},
  journal={IEEE Access},
  volume={12},
  pages={176554--176568},
  year={2024},
  publisher={IEEE}
}