OFDM PHY Signal Processing Project

End-to-end physical layer implementation — C++17 + Python 3.11

Built as a portfolio project demonstrating PHY software engineering skills relevant to 4G/5G base station development.

---

What's in this repo

├── cpp/
│   ├── dsp.h          # Cooley-Tukey Radix-2 FFT/IFFT from scratch
│   ├── ofdm.h         # OFDM modulator/demodulator (LTE numerology)
│   ├── channel.h      # AWGN + Rayleigh fading channel models
│   └── main.cpp       # BER vs SNR simulation (outputs ber_results.csv)
│
├── python/
│   └── spectrum_analyzer.py   # RF spectrum analysis + signal detection
│
└── docs/
    └── project_writeup.docx   # Full technical write-up

---

Component 1 — C++ OFDM PHY Simulator

Full LTE baseband chain, no external libraries:

Random bits → QAM mapper → Subcarrier mapping → IFFT → Cyclic prefix → TX signal
                                                                              ↓
                                                                       AWGN channel
                                                                              ↓
Decoded bits ← QAM demapper ← FFT ← CP removal ← RX signal

Build & run

g++ -O2 -std=c++17 -o ofdm_ber cpp/main.cpp
./ofdm_ber

Output: BER table printed to stdout + ber_results.csv for plotting.

Key parameters (3GPP TS 36.211)

Parameter	Value
FFT size	128
Active subcarriers	72 (6 RBs)
Subcarrier spacing	15 kHz
Cyclic prefix	9 samples (normal CP)
Modulation	BPSK / QPSK / 16-QAM / 64-QAM

FFT self-test

IFFT(FFT(x)) max error: 5.2e-16  ✓ (floating-point machine epsilon)
Zero-noise BER: 0 bit errors across all modulation orders  ✓

---

Component 2 — Python Spectrum Analyser

Processes real RTL-SDR IQ captures or synthetic signals.

Install

pip install numpy scipy matplotlib

Usage

# Synthetic demo (no hardware needed)
python python/spectrum_analyzer.py --demo --fc 1800 --fs 20.0

# Real RTL-SDR capture
python python/spectrum_analyzer.py --file capture.bin --fc 1800 --fs 2.0

# Plot BER curves from C++ output
python python/spectrum_analyzer.py --ber ber_results.csv

Pipeline

IQ samples → Hanning window → FFT (Welch PSD) → dBm conversion
          → Peak detection → 3dB BW estimation → Signal classification
          → Spectrum plot + Waterfall display

Signal classification (bandwidth heuristic)

Occupied BW	Classification
≥ 80 MHz	5G NR (100 MHz)
18–25 MHz	LTE 20 MHz
8–13 MHz	LTE 10 MHz
4–8 MHz	LTE 5 MHz

---

RTL-SDR hardware setup (when available)

# Install drivers
sudo apt install rtl-sdr gnuradio

# Test dongle
rtl_test -t

# Capture LTE Band 3 (1800 MHz)
rtl_sdr -f 1800000000 -s 2000000 -n 2000000 capture.bin

# Analyse
python python/spectrum_analyzer.py --file capture.bin --fc 1800 --fs 2.0

---

Standards references

• 3GPP TS 36.211 — LTE Physical channels and modulation
• 3GPP TS 38.211 — NR Physical channels and modulation
• Cooley & Tukey (1965) — "An Algorithm for the Machine Calculation of Complex Fourier Series"

---

Planned extensions

• Frequency-domain channel equaliser (ZF / MMSE)
• LDPC/Turbo FEC to show coding gain on BER curves
• SIMD optimisation using Intel AVX2 intrinsics
• LTE cell search (PSS/SSS detection)
• GNU Radio real-time integration