SISO OFDM radar processing chain that recovers range and velocity of multiple moving targets from a single simulated transmit/receive cycle, using matrix-based subcarrier processing instead of matched filtering.
Classical pulsed radar dedicates its waveform to ranging. OFDM radar reuses a communications-style multicarrier waveform (as in automotive 77 GHz systems) for joint range-velocity sensing, extracting both from the same transmitted symbols via subcarrier-domain matrix operations — no matched filter bank required.
Transmit OFDM symbols S (Nc subcarriers × Nsym symbols)
│
▼
Per-target channel model Range → per-subcarrier phase rotation
Velocity → inter-carrier interference (ICI) + common phase error (CPE)
│
▼
Received signal Y = Σ_targets a · D_ICI · IDFT(D_phase · S) · D_CPE
│
▼
Element-wise division by S Removes known transmit data, isolates channel response
│
▼
IDFT along subcarriers Range dimension
│
▼
DFT along symbols Doppler (velocity) dimension
│
▼
Range-Doppler Map (RDM)
Two synthetic targets recovered from a simulated 1024-subcarrier, 256-symbol, 1 GHz-bandwidth, 77 GHz-carrier link:
| Target | Range | Velocity |
|---|---|---|
| 1 | 30.5 m | −30 m/s |
| 2 | 90.5 m | 110 m/s |
Theoretical resolution at these parameters: ΔR = c/(2B) = 0.15 m, Δv = λ/(2·N_sym·T) ≈ 1.46 m/s.
├── OFDM.py # QPSK subcarrier symbol generation
├── OFDM_Radar_System.py # Main radar processing chain (class-based)
├── ofdm_test.py # Alternate vectorized implementation (scipy.fft-based)
├── plotRDM.py / plotRDM.m # Standalone RDM plotting utilities (Python / MATLAB)
└── RDM.png # Example output
OFDM_Radar_System.py and ofdm_test.py implement the same chain two ways — the former mirrors the matrix-algebra derivation directly (explicit DFT matrices), the latter uses scipy.fft for the same operations. Kept both since they're a useful side-by-side reference for the two implementation styles.
pip install numpy scipy matplotlib
python OFDM_Radar_System.pyEdit the objects list in __main__ to place targets at different range/velocity pairs.
| Component | Method | Why |
|---|---|---|
| Range | Phase rotation per subcarrier, recovered via IDFT | Time delay in a multicarrier signal is a linear phase ramp across subcarriers |
| Velocity | ICI + CPE matrices, recovered via DFT across symbols | Doppler shift couples subcarriers (ICI) and drifts phase symbol-to-symbol (CPE) |
| Channel removal | Element-wise division by known transmit symbols | Standard OFDM radar trick — no correlation/matched filter needed since the waveform is already known |
- Single transmit/receive antenna (SISO) — no angular (DoA) information
- No noise model — targets are recovered from a clean simulated channel
- ICI/CPE model assumes constant velocity within the coherent processing interval
