⚡ electrical engineering

A Novel One-tap Equalizer for Zero-Padded AFDM System over Doubly Selective Channels

This paper proposes a novel low-complexity one-tap equalizer for zero-padded AFDM systems over doubly selective channels, which simplifies the input-output relation through specific parameter selection and zero-padding to enable efficient symbol recovery in a newly defined frequency-of-affine (FoA) domain.

Original authors: Chenyang Zhang, Akram Shafie, Cheng Shen, Deepak Mishra, Jinhong Yuan

Published 2026-03-04

📖 6 min read🧠 Deep dive

CC BY 4.0

Original authors: Chenyang Zhang, Akram Shafie, Cheng Shen, Deepak Mishra, Jinhong Yuan

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

The Big Picture: The "Moving Train" Problem

Imagine you are trying to send a message to a friend while you are both on a speeding train. The train is shaking, the wind is howling, and the tracks are bumpy. In the world of wireless communication, this is called a "doubly selective channel." It means the signal is changing because of time (the train is moving fast) and frequency (the Doppler effect, like a siren passing by).

Traditional systems (like the Wi-Fi in your home or 4G/5G phones) are like sending a postcard through a calm river. They work great when the water is still. But when the river turns into a raging, twisting torrent (high-speed trains, drones, or underwater communication), the postcards get scrambled, torn, and lost.

The authors of this paper are trying to build a new way to send messages that can survive this "raging river" without needing a super-complex, expensive, and slow computer at the receiving end to fix the mess.

The Old Way: The "Jigsaw Puzzle" Nightmare

The current best technology for these fast-moving scenarios is called AFDM (Affine Frequency Division Multiplexing). Think of AFDM as a very clever way of packing your message into a box so it doesn't break.

However, when the message arrives at the other end, it's like a jigsaw puzzle that has been shaken in a box. The pieces are all mixed up. To fix it, the receiver usually needs a "Multi-Tap Equalizer."

The Metaphor: Imagine you have a tangled ball of 1,000 headphones. To untangle them, you have to pull on every single knot, one by one, checking how they connect to their neighbors. This takes a lot of time and energy (computational power). It's accurate, but it's slow and expensive.

The New Idea: The "Magic Box" with a One-Tap Fix

The authors propose a new system called ZP-AFDM (Zero-Padded AFDM) with a One-Tap Equalizer. They want to turn that tangled ball of headphones into a neat, straight line that you can fix with a single tug.

Here is how they do it, step-by-step:

1. Tuning the Radio (The Parameters $c_1$ and $c_2$ )

In the old system, the "knobs" on the radio (parameters $c_1$ and $c_2$ ) were set to standard values. The authors say, "Let's turn the $c_1$ knob way up high."

The Analogy: Imagine you are trying to hear a whisper in a noisy room. Usually, you just shout louder. But here, the authors say, "Let's change the pitch of the whisper so high that the background noise can't touch it." By turning up this specific parameter, the chaotic mess of the signal becomes much more orderly. The "shaking" of the signal slows down significantly.

2. The "Zero-Padding" Trick (Leaving Empty Seats)

To make the signal even easier to fix, they add "Zero Padding."

The Analogy: Imagine a theater where the audience is rowdy and keeps bumping into each other. The authors say, "Let's leave the first and last few rows of seats completely empty."
Why? In the old system, the "bumping" (interference) happened at the edges of the message, ruining the data. By leaving empty seats (zeros) at the start and end, any "bumping" that happens there just hits empty space and disappears. It acts like a safety buffer.

3. The "Cyclic Reconstruction" (The Magic Wrap)

At the receiving end, they do a clever trick called "Cyclically Superimposed Reconstruction."

The Analogy: Imagine you have a long strip of paper with a message, but the edges are frayed. Instead of throwing away the frayed parts, you take the frayed end and tape it to the beginning of the strip, making a loop.
The Result: This turns the messy, linear "tangled headphones" into a perfect circle. In math, a circle is much easier to work with than a straight line because it has no "ends" to get confused.

4. The "Frequency-of-Affine" (FoA) Domain (The New Language)

Now that the signal is a neat circle, they translate it into a new language called the FoA domain.

The Analogy: Imagine you have a song that sounds like static noise. You put it through a special filter (the Fourier Transform) that separates the static into individual notes. Suddenly, you see that the song is actually just one clear melody with a tiny bit of background hiss.
The Magic: In this new "FoA" language, the complex, tangled mess of the channel looks like a straight diagonal line. Every piece of data only talks to its own matching piece. They don't talk to their neighbors anymore!

5. The "One-Tap Equalizer" (The Single Fix)

Because the signal is now a neat diagonal line in the FoA domain, the receiver doesn't need to untangle 1,000 knots.

The Analogy: Instead of a master plumber fixing a whole house of pipes, you just need a single wrench to tighten one bolt.
The Result: The receiver looks at each piece of data, applies a simple math correction (one "tap"), and the message is recovered perfectly. This is incredibly fast and uses very little battery power.

The Trade-Off: Efficiency vs. Simplicity

The paper admits there is a small price to pay. Because they had to leave those "empty seats" (Zero Padding) at the start and end of the message, they can't send quite as many actual data bits in the same amount of time.

The Metaphor: It's like driving a car with a large, heavy bumper. It protects you better in a crash (high speed), but it makes the car slightly heavier and less fuel-efficient.
The Verdict: The authors show that even with this small loss in speed, the system is much better than the old methods. It handles high speeds (like 500 km/h) without errors, whereas the old methods (like standard OFDM) would fail completely.

Summary

This paper presents a clever engineering trick to make wireless communication robust against high-speed movement.

Tune the knobs to slow down the signal chaos.
Add empty space (zeros) to protect the edges.
Wrap the signal into a loop to make it mathematically simple.
Translate it into a new language where the mess disappears.
Fix it with a single, simple tool (One-Tap Equalizer).

The result is a system that is fast, simple, and tough, perfect for the future of high-speed trains, drones, and underwater communication.

1. Problem Statement

Context: Orthogonal Frequency Division Multiplexing (OFDM), the backbone of 4G/5G, struggles in doubly selective channels (channels that vary in both time and frequency, such as high-mobility or underwater acoustic scenarios). Severe Doppler spread destroys subcarrier orthogonality, leading to Inter-Carrier Interference (ICI).
Existing Solutions:

Delay-Doppler (DD) Modulation: Schemes like OTFS and ODDM offer robustness but often require complex receivers.
Affine Frequency Division Multiplexing (AFDM): A newer modulation scheme that multiplexes symbols in the "affine domain." While it offers performance comparable to OTFS, existing equalization methods for AFDM (e.g., Decision Feedback Equalizers, Message Passing Algorithms, or Multi-tap MMSE) are computationally intensive.
The Gap: There is a lack of a low-complexity, one-tap equalizer for AFDM that rivals the simplicity of OFDM's frequency-domain equalization while maintaining robustness against high Doppler shifts.

2. Methodology

The authors propose a Zero-Padded AFDM (ZP-AFDM) system coupled with a novel Frequency-of-Affine (FoA) domain one-tap equalizer. The methodology involves four key steps:

A. Strategic Parameter Selection ( $c_1$ and $c_2$ )

Standard AFDM uses parameters $c_1$ (post-chirp) and $c_2$ (pre-chirp).
The authors select $c_1$ to be significantly larger than the minimum required for diversity ( $c_1 = \chi \frac{2k_{max}+1}{2N}$ , where $\chi > 1$ ).
They set $c_2 = \frac{1}{4c_1N^2}$ , satisfying the condition $4c_1c_2N^2 = 1$ .
Result: This specific selection simplifies the Input-Output Relation (IOR) in the affine domain. It reveals that the phase variation experienced by affine domain symbols is significantly slower than that of time-domain symbols. Specifically, the effective frequency shift becomes $\bar{f}_{i,aff} = \frac{k_i}{2c_1N}$ , which is much smaller than the time-domain shift.

B. Zero-Padding and Signal Reconstruction

Transmitter: Zero-padding is applied to the transmitted affine domain symbols. Specifically, $L_z$ symbols at the beginning and end of the block are set to zero. This mitigates the impact of "additional phase terms" ( $D'_i[m]$ ) that occur at the boundaries of the received symbol vector due to channel delays and Doppler shifts.
Receiver: A cyclically superimposed reconstruction is performed. The last $L_z$ received symbols are added to the first $L_z$ symbols.
Result: This transformation converts the effective affine domain IOR from a linear convolution with boundary effects into an approximately circular convolution characterized by a slowly varying phase term.

C. Transformation to Frequency-of-Affine (FoA) Domain

The reconstructed time-domain-like affine symbols are transformed using a Discrete Fourier Transform (DFT) into a new domain called the Frequency-of-Affine (FoA) domain.
Key Insight: In the FoA domain, the effective channel matrix becomes nearly diagonal. The off-diagonal elements (representing interference between subcarriers) are suppressed because the effective frequency shifts are small (due to the large $c_1$ ) and the sinc-like nature of the channel kernel limits energy leakage to adjacent subcarriers.

D. One-Tap Equalization

Because the channel matrix in the FoA domain is nearly diagonal, the authors propose a simple one-tap equalizer (similar to OFDM).
The equalizer coefficient $E[k]$ is calculated based on the diagonal channel element and the total noise-plus-interference power.
The equalized symbols are transformed back to the affine domain via IDFT for symbol detection.

3. Key Contributions

Novel Parameter Selection: Demonstrating that increasing the chirp parameter $c_1$ (beyond the diversity minimum) drastically reduces the effective Doppler spread in the affine domain, enabling simpler equalization.
ZP-AFDM Framework: Introducing a specific zero-padding and cyclic reconstruction scheme that eliminates boundary phase distortions and enables circular convolution properties in the affine domain.
FoA Domain Concept: Defining the "Frequency-of-Affine" domain where the channel matrix is diagonally dominant, allowing for the application of low-complexity one-tap equalization.
Low-Complexity Receiver: Achieving AFDM performance with a receiver complexity comparable to OFDM (one-tap), a significant reduction compared to existing multi-tap or iterative AFDM equalizers.

4. Numerical Results

The authors evaluated the scheme using the EVA channel model with a maximum UE speed of 500 km/h (high mobility).

BER Performance:
- With a high $\chi$ (e.g., $\chi=17$ ), the proposed ZP-AFDM with one-tap equalizer achieves Bit Error Rate (BER) performance comparable to AFDM with complex multi-tap MRC equalizers.
- It significantly outperforms standard OFDM and Single-Carrier Frequency Domain Equalization (SC-FDE) with one-tap equalizers in the same doubly selective environment.
- Even with lower $\chi$ (e.g., $\chi=9$ ), the scheme outperforms OFDM/SC-FDE, though an error floor appears at very low BER ( $10^{-6}$ ).
Trade-off (Spectral Efficiency vs. Performance):
- Increasing $\chi$ (increasing $c_1$ ) improves BER by narrowing the channel matrix band but increases the zero-padding overhead, reducing spectral efficiency.
- The paper suggests an optimal $\chi$ exists (e.g., $\chi=13$ at 22 dB SNR) to balance BER and spectral efficiency.
Robustness: The scheme maintains full diversity and shows no error floor up to $10^{-6}$ BER with 10% overhead when $\chi$ is sufficiently high.

5. Significance

This work bridges the gap between the robustness of advanced modulation schemes (like AFDM/OTFS) and the implementation simplicity of OFDM.

High-Mobility Viability: It provides a viable solution for future high-speed communication systems (e.g., 6G, high-speed trains, UAVs) where Doppler shifts are severe.
Complexity Reduction: By enabling a one-tap equalizer, the computational burden at the receiver is drastically reduced, making the technology more feasible for real-time hardware implementation compared to iterative or multi-tap alternatives.
Theoretical Insight: It offers a new perspective on manipulating the affine domain parameters to control the "effective" Doppler spread, transforming a complex time-variant channel problem into a quasi-static frequency-domain problem.