This project dives into Rayleigh fading channels for Non-Orthogonal Multiple Access (NOMA) users. π‘ Leveraging Monte Carlo simulations, it explores signal detection across different Signal-to-Noise Ratios (SNRs). π Plus, we harness the power of Machine Learning to boost signal detection capabilities.
- numpy: For number crunching π’.
- pandas: Expert at data handling and manipulation π.
- matplotlib: Our go-to for stunning visualizations and plots π.
- scipy: A powerhouse for scientific computing, essential for interpolation π¬.
- scikit-learn: The brain behind our Machine Learning models π€.
- Monte Carlo iterations:
num_iter = 10000
- NOMA users:
N = 2
- Power allocations:
a1, a2 = 0.8, 0.2
- Sampled length:
S = 4096
- False-alarm probability:
Pf = 0.1
- Environmental SNR range: -25dB to 5dB π.
- Transmitter power:
transmitter_power = 1
Utilizing the Monte Carlo simulation, we:
- Craft NOMA signals with random cyclic delays π.
- Merge these to form our transmitted signal πΆ.
- Stir in Rayleigh distributed noise for realism πͺοΈ.
- Detect NOMA signals with cyclic correlation π.
We train three ML champions:
- Logistic Regression (LR)
- Random Forest (RF)
- Decision Tree (DT)
They're on a mission to spot NOMA signals. We assess their prowess using the ROC curve, eyeing the True Positive Rate (TPR) against a False Positive Rate (FPR) of 0.1.
We plot Detection Probability vs. Environmental SNR, showcasing:
- Classic signal detection (sans ML) π¦
- Logistic Regression (LR) π
- Random Forest (RF) π³
- Decision Tree (DT) π²
- Clone our repository using
git clone https://github.com/suyashvsingh/NOMA-ML-Spectrum-Detection-CIoT.git
π. - Install the dependencies using
pip install -r requirements.txt
π¦. - Fire up the notebook for an epic plot showdown: traditional vs ML-augmented detection under varied SNRs π.
A stunning graph, "Probability of Detection vs. Environmental SNR," awaits you. It pits traditional detection against our ML trio, LR, RF, and DT, focusing on TPR at a steady FPR of 0.1.
Jump into:
- Turbocharge the Monte Carlo simulation βοΈ.
- Bring in cutting-edge ML algorithms π§ .
- Elevate our visualization game π¨.
Note: For the whole story and deep insights, explore the notebook π.