- Jupyter Notebook 96%
- Python 4%
| Filename | Latest commit message | Latest commit date |
|---|---|---|
| .gitignore | ||
| Channel_Estimation_in_OTFS_and_NOMA_.ipynb | ||
| channel_estimation_in_otfs_and_noma_.py | ||
| config.py | ||
| LICENSE | ||
| main_pipeline.py | ||
| README.md | ||
| requirements.txt | ||
| utils.py | ||
📡 Channel Estimation in OTFS and NOMA using DBGP-CS
📖 Overview This repository presents a hybrid Deep Bayesian Gaussian Process with Compressive Sensing (DBGP-CS) framework for channel estimation (CE) in OTFS-NOMA wireless systems.
The proposed approach integrates:
• Deep Neural Networks (DNN) for feature extraction
• Gaussian Processes (GP) for probabilistic modeling
• Compressive Sensing (CS) for sparse recovery
It is designed for high-mobility environments with severe:
• Doppler shifts
• Multipath propagation
• Interference
🎯 Key Contributions
✔ Hybrid DBGP-CS model for OTFS-NOMA CE
✔ 50% reduction in pilot overhead
✔ Improved:
o NMSE (Normalized Mean Squared Error)
o BER (Bit Error Rate)
✔ Robust under:
o High mobility (up to 120 km/h)
o Dense multipath channels
✔ Supports 5G/6G vehicular communication systems
🧠 Methodology
🔁 System Pipeline
OTFS Modulation → NOMA Superposition → Channel (DD domain)
→ Noise Addition → OTFS Demodulation
→ Feature Extraction (DNN)
→ Sparse Recovery (CS - OMP)
→ Channel Estimation (Gaussian Process)
→ Performance Evaluation (NMSE, BER)
⚙️ Implementation Details
Component Technique
Modulation OTFS (ISFFT/SFFT)
Multiple Access NOMA
Feature Extraction Deep Neural Network
Estimation Gaussian Process
Sparse Recovery Orthogonal Matching Pursuit
Metrics NMSE, BER
📊 Experimental Setup
Parameter Value
Carrier Frequency 28 GHz
Modulation 16-QAM
Users 100
Mobility 60–120 km/h
SNR Range 0–12 dB
Channel Model Sparse Delay-Doppler
📈 Results
The model is evaluated using:
• NMSE vs SNR
• BER vs SNR
• Mobility impact
• Pilot overhead analysis
• Energy allocation
• Multipath complexity
📊 Generated Figures
Figure Description
Fig 3 NMSE vs SNR
Fig 4 BER vs SNR
Fig 5 NMSE vs Mobility
Fig 6 BER vs Mobility
Fig 7 BER vs Pilot Count
Fig 8 NMSE vs Pilot Count
Fig 9 NMSE vs Energy Allocation
Fig 10 BER vs Energy Allocation
Fig 11 NMSE vs Multipath
Fig 12 BER vs Multipath
📁 Repository Structure
.
├── main_pipeline.py # End-to-end execution
├── config.py # Hyperparameters
├── utils.py # Helper functions
├── requirements.txt # Dependencies
├── README.md
├── LICENSE
🚀 Installation
git clone https://github.com/yourusername/Channel-Estimation-in-OTFS-and-NOMA.git
cd Channel-Estimation-in-OTFS-and-NOMA
pip install -r requirements.txt
▶️ Usage
python main_pipeline.py
📌 Example Output
SNR 0 dB → NMSE: 0.2446, BER: 0.7782
SNR 12 dB → NMSE: 0.0144, BER: 0.0211
⚠️ Limitations
• Simplified channel model (can extend to 3GPP ETU)
• Approximate Gaussian Process implementation
• No LDPC coding (can be extended)
• Offline training required
🔮 Future Work • Integration with LDPC / Polar codes
• Massive MIMO-OTFS extension
• Real-time edge deployment
• Model compression (pruning, quantization)
• Full Bayesian GP (GPyTorch)
📄 License
This project is licensed under the MIT License.
👩💻 Author
Nitha Anilkuma, Sudhakar Sengan Department of Computer Science and Engineering, PSN College of Engineering and Technology, Tirunelveli, 627451, Tamil Nadu, India.