MATLAB Simulation of a High-Mobility AFDM Waveform for Joint Communication and Radar Sensing
This project implements and evaluates an Affine Frequency Division Multiplexing (AFDM)-based Integrated Sensing and Communication (ISAC) system in MATLAB.
The objective is to demonstrate how a single AFDM waveform can support both:
- wireless communication
- radar target sensing
within a high-mobility, delay-Doppler environment.
The project evaluates both communication and sensing performance using a complete simulation chain that includes AFDM modulation/demodulation, doubly-selective channels, pilot-aided channel estimation, equalization, BER analysis, Range-Doppler processing, and CFAR-based target detection. The project materials describe the motivation around Doppler sensitivity and the use of AFDM for simultaneous communication and sensing.
High mobility introduces significant Doppler effects that can degrade conventional OFDM-based communication and sensing systems.
AFDM is investigated as a waveform designed to better handle delay-Doppler dispersion and high-mobility channels.
High Mobility
│
▼
Delay + Doppler
│
▼
Conventional OFDM
Performance
Degradation
│
▼
AFDM
│
┌─────────┴─────────┐
▼ ▼
Communication Sensing
│ │
└─────────┬─────────┘
▼
ISAC
The project follows the AFDM-ISAC approach described in the provided project materials, where one waveform serves both communication and sensing functions.
The implemented simulation follows this processing chain:
Information Bits
│
▼
16-QAM
│
▼
Pilot Insertion
│
▼
IDAFT
│
▼
Chirp-Periodic Prefix
│
▼
Doubly-Selective Channel
│
├───────────────┐
│ │
▼ ▼
Communication Radar Echo
Receiver Model
│ │
▼ ▼
DAFT Range-Doppler
│ Processing
▼ │
Pilot-Aided LS │
Estimation │
│ │
▼ ▼
ZF Equalization FCCR / DAFT
│ │
▼ ▼
BER / SER / EVM Range + Velocity
│
▼
CFAR
│
▼
Target Detection
The transmitter consists of:
Random Bits
↓
QAM Mapping
↓
Pilot Insertion
↓
IDAFT / AFDM Modulation
↓
Chirp-Periodic Prefix
↓
Transmitted AFDM Waveform
The current implementation uses the following main simulation parameters:
| Parameter | Value |
|---|---|
| Subcarriers | 512 |
| AFDM symbols | 32 |
| Modulation | 16-QAM |
| Main SNR | 15 dB |
| SNR sweep | -5 to 30 dB |
| Monte-Carlo trials | 20 |
| Carrier frequency | 24 GHz |
| Subcarrier spacing | 22.729 kHz |
| Chirp-Periodic Prefix | 64 |
| Pilot spacing | 1 pilot every 8 subcarriers |
| Radar targets | 3 |
| Target ranges | 300 m, 200 m, 120 m |
| Target velocities | 20 m/s, 8 m/s, -8 m/s |
These values are taken from the implemented MATLAB simulation rather than invented for the README.
The communication path includes:
16-QAM
↓
Pilot Insertion
↓
AFDM Modulation
↓
Doubly-Selective Communication Channel
↓
DAFT Demodulation
↓
Pilot-Based LS Channel Estimation
↓
Zero-Forcing Equalization
↓
QAM Demodulation
↓
BER / SER / EVM
The MATLAB implementation performs pilot-aided least-squares channel estimation and ZF equalization before computing communication metrics.
Known pilot symbols are inserted using a comb-type pilot pattern across the AFDM frame.
The implemented configuration uses one pilot every 8 subcarriers, giving a 12.5% pilot overhead.
The radar path models multiple targets with different delays and Doppler shifts.
AFDM Waveform
│
▼
Target Delay
+
Target Doppler
│
▼
Radar Echo
│
▼
FCCR Processing
│
▼
Range-Doppler Map
A second sensing path operates in the DAFT domain:
AFDM Echo
↓
DAFT Domain
↓
DAFT-Based Sensing
↓
Range-Doppler Map
↓
CFAR Detection
The project source implements both time-domain FCCR sensing and DAFT-domain sensing, followed by CFAR detection.
The simulation models three radar targets:
Target 1 → Range = 300 m | Velocity = +20 m/s
Target 2 → Range = 200 m | Velocity = +8 m/s
Target 3 → Range = 120 m | Velocity = -8 m/s
The source explicitly defines these target locations and velocities and converts them into propagation delays and Doppler frequencies.
The project evaluates AFDM communication performance over an SNR sweep and compares AFDM with baseline waveforms.
The implementation performs Monte-Carlo BER evaluation and also generates theoretical AWGN reference results.
The simulation evaluates communication robustness against increasing normalized Doppler and compares AFDM with OFDM.
The sweep evaluates normalized Doppler from 0 to 2 at a fixed SNR.
The transmitted and recovered QAM symbol distributions are compared after channel estimation and equalization.
The project evaluates the difference between the known target range/velocity values and the values estimated by the sensing processing chain.
The sensing image quality is evaluated as the input SNR varies.
The comparison evaluates the image SNR obtained from the AFDM-FCCR and AFDM-DAFT-domain sensing approaches.
A Cell-Averaging CFAR (CA-CFAR) detector is applied to the DAFT-domain Range-Doppler response.
Range-Doppler Map
│
▼
Power Map
│
▼
Training Cells
+
Guard Cells
│
▼
Adaptive Threshold
│
▼
Target Detection
The current implementation uses:
- Probability of false alarm:
1e-3 - Training cells:
8 - Guard cells:
2
and extracts detected target range and velocity from connected detection regions.
The simulation evaluates both communication and sensing metrics.
- BER
- SER
- EVM
- Spectral efficiency
- BER vs SNR
- BER vs normalized Doppler
- Range estimation error
- Velocity estimation error
- PSLR
- Range-Doppler Maps
- CFAR detections
- Doppler ambiguity / velocity range
The source prints these metrics as part of the final simulation summary.
MATLAB
Signal Processing
Wireless Communications
Radar Signal Processing
AFDM
IDAFT / DAFT
16-QAM
Pilot-Based Channel Estimation
Zero-Forcing Equalization
Range-Doppler Processing
CFAR Detection
Monte-Carlo Simulation
AFDM-ISAC-Simulation/
│
├── src/
│ └── AFDM_ISAC_Simulation.m
│
├── results/
│ ├── communication/
│ │ ├── ber-vs-snr.png
│ │ ├── ber-vs-doppler.png
│ │ ├── constellation.png
│ │ └── pilot-structure.png
│ │
│ ├── sensing/
│ │ ├── estimation-errors.png
│ │ ├── image-snr-vs-input-snr.png
│ │ ├── fccr-rdm-2d.png
│ │ ├── fccr-rdm-3d.png
│ │ ├── daft-rdm-2d.png
│ │ └── daft-rdm-3d.png
│ │
│ └── detection/
│
├── docs/
│ └── architecture/
│ ├── system-overview.png
│ └── transmitter-chain.png
│
└── README.md
- Open MATLAB.
- Set the project directory as the MATLAB working directory.
- Open:
src/AFDM_ISAC_Simulation.m
- Run the script.
- Inspect the generated communication and sensing figures.
The simulation produces communication performance metrics, sensing maps, estimation-error plots, and detection results.
This project demonstrates practical work across multiple areas of communication and sensing engineering:
Wireless Communication
│
├── Modulation
├── Channel Modeling
├── Channel Estimation
└── Equalization
│
▼
Signal Processing
│
┌───────┴────────┐
▼ ▼
Radar Communication
Sensing Metrics
│ │
▼ ▼
Range / Velocity BER / SER / EVM
Estimation
│
▼
CFAR Detection
Potential extensions include:
- More realistic multi-target and channel models
- Larger Monte-Carlo simulations
- More extensive AFDM vs OFDM vs OTFS comparisons
- Improved channel-estimation methods
- Advanced CFAR variants
- Joint communication-sensing optimization
- Real-time or software-defined-radio implementation
- Automotive radar-oriented scenario modeling
Amit Tanwar
Electronics and Communication Engineering
This project investigates AFDM as an ISAC waveform for high-mobility communication and radar sensing.
The implementation combines:
AFDM
+
16-QAM
+
Pilot-Based Channel Estimation
+
Doubly-Selective Channels
+
BER / EVM Analysis
+
FCCR Sensing
+
DAFT-Domain Sensing
+
Range-Doppler Maps
+
CA-CFAR Detection
The result is a MATLAB-based simulation framework that demonstrates how communication and sensing can be evaluated together within a single AFDM-based system.