🔢 mathematics

Non-Orthogonal Affine Frequency Division Multiplexing for Spectrally Efficient High-Mobility Communications

This paper proposes a novel non-orthogonal affine frequency division multiplexing (nAFDM) waveform that enhances spectral efficiency for high-mobility communications by introducing controllable subcarrier overlapping via a bandwidth compression factor, supported by an efficient IDFT-based implementation and a soft iterative detection algorithm to mitigate inter-carrier interference while maintaining robust bit error rate performance.

Original authors: Qin Yi, Zilong Liu, Leila Musavian, Zeping Sui

Published 2026-02-17

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Qin Yi, Zilong Liu, Leila Musavian, Zeping Sui

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

Imagine you are trying to send a series of postcards (data) to a friend while riding a rollercoaster at top speed. The wind is howling, the track is shaking, and your friend is moving just as fast. This is the challenge of high-mobility communication (like in self-driving cars, high-speed trains, or drones).

In the old days, we used a method called OFDM (Orthogonal Frequency Division Multiplexing). Think of this like sending postcards on separate, perfectly spaced lanes of a highway. As long as the lanes stay straight, everyone arrives safely. But on a rollercoaster (high speed), the lanes get twisted and bent. The postcards start crashing into each other, causing a mess called Inter-Carrier Interference (ICI).

Recently, a new method called AFDM was invented. Instead of straight lanes, AFDM uses "chirp" lanes—like a siren sound that changes pitch. These lanes are flexible and can bend with the rollercoaster, keeping the postcards safe even when things get bumpy.

The Problem:
While AFDM is great at handling the bumps, it's a bit wasteful with space. It leaves gaps between the lanes to ensure they never touch. In a world where we need to send more data than ever, those gaps are a luxury we can't afford.

The Solution: nAFDM (Non-Orthogonal AFDM)
This paper proposes a new, clever twist called nAFDM.

The Core Idea: The "Crowded Elevator" Analogy

Imagine you have an elevator (the bandwidth) that can hold 10 people comfortably if they stand in perfect, spaced-out rows (Orthogonal AFDM).

The authors ask: "What if we squeeze everyone in a little tighter?"

They introduce a "Bandwidth Compression Factor." This is like telling everyone to stand shoulder-to-shoulder instead of leaving a foot of space.

The Benefit: You can fit 20% more people (data) into the same elevator. This is Spectral Efficiency (SE).
The Risk: Because people are so close, they bump into each other. In radio terms, this is Interference. If you just squeeze them in without a plan, the elevator becomes a chaotic mess, and no one gets their message across.

How They Solved the Chaos

The paper doesn't just squeeze the elevator; it invents a new way to manage the crowd so they don't crash.

The "Smart Squeeze" (IDFT Generation):
Usually, squeezing people in requires building a whole new, expensive elevator (hardware). The authors found a mathematical trick (using existing IDFT/IFFT modules) to squeeze the signal in without needing new hardware. It's like rearranging the furniture in your living room to fit a bigger sofa without buying a new house.
The "Soft Detective" (Soft Iterative Detection):
When the signal arrives at the receiver, it's a jumbled mess because of the bumps and the crowding.
- Old Way (Linear Detection): Imagine a detective who looks at a blurry photo and guesses, "That's a cat." If they are wrong, they stop there.
- New Way (Soft Iterative Detection): The authors created a "Super Detective." This detective looks at the blurry photo, makes a guess, then says, "Hmm, that looks a bit like a dog, but maybe it's a cat. Let me check the shadows again."
- They use probability (soft information). Instead of making a hard decision immediately, they calculate the likelihood of what the data is. They then use that guess to cancel out the interference from neighbors, refine the guess, and repeat the process. It's like cleaning a muddy window: you wipe a little, look again, wipe a little more, until the picture is crystal clear.
The "Selective Cleanup" (ICI Pruning):
Sometimes, the "Super Detective" gets overwhelmed trying to check every single neighbor. The paper suggests a shortcut: "Only worry about the neighbors who are standing right next to you; the ones across the room aren't bothering you much."
By ignoring the tiny, distant bumps and focusing only on the big, immediate ones, they save a lot of computing power (battery life) while still getting a clear picture.

The Results: Why Should We Care?

The simulations in the paper show that this new nAFDM system is a game-changer:

It's Faster: It packs more data into the same space than current standards (like OFDM).
It's Tougher: It handles high-speed movement (like a train going 200 mph) just as well as the current best technology (AFDM).
It's Smart: Even though the data is packed tightly, the "Soft Detective" algorithm cleans up the mess so well that the error rate is almost the same as if the data were spaced out loosely.

In Summary:
The authors took a robust system (AFDM) that works well on bumpy roads, figured out how to pack it tighter to carry more cargo, and built a smart, iterative cleaning crew (the algorithm) to sort out the inevitable mess caused by the tight packing. The result is a communication system that is faster, more efficient, and ready for the high-speed future.

1. Problem Statement

High-mobility communication scenarios (e.g., high-speed rail, UAVs, V2X) present significant challenges due to doubly selective channels (time-varying and frequency-selective), which cause severe Inter-Carrier Interference (ICI) and degrade the performance of traditional Orthogonal Frequency Division Multiplexing (OFDM).

Current Solutions: Affine Frequency Division Multiplexing (AFDM) has emerged as a robust alternative to OFDM and OTFS, utilizing chirp subcarriers to handle Doppler shifts effectively. However, conventional AFDM relies on orthogonal subcarriers, limiting its Spectral Efficiency (SE).
The Gap: While non-orthogonal waveforms like Spectrally Efficient Frequency Division Multiplexing (SEFDM) exist to boost SE by overlapping subcarriers, they are highly sensitive to Doppler shifts and suffer from performance degradation in high-mobility environments.
Objective: The paper aims to design a waveform that combines the high SE of non-orthogonal signaling with the robustness of AFDM in high-mobility scenarios, while providing efficient detection algorithms to mitigate the resulting interference.

2. Methodology

A. System Model: Non-Orthogonal AFDM (nAFDM)

The authors propose nAFDM, a novel waveform that introduces a bandwidth compression factor ( $\alpha$ ) into the AFDM modulator.

Modulation: Instead of strictly orthogonal subcarriers ( $\alpha=1$ ), nAFDM uses $\alpha \in (0, 1)$ . This compresses the subcarrier spacing, allowing them to overlap and increasing the number of symbols transmitted per unit bandwidth.
Signal Generation: To avoid the high computational cost of direct implementation, the authors propose an efficient Inverse Discrete Fourier Transform (IDFT)-based generation method. By zero-padding the input symbols and using an extended IFFT of length $N' = N/\alpha$ , the system generates the nAFDM waveform using existing hardware modules with complexity comparable to standard OFDM/AFDM.
Channel Characteristics: The introduction of $\alpha < 1$ breaks the chirp periodicity, necessitating a standard Cyclic Prefix (CP) rather than the Chirp-Periodic Prefix (CPP) used in orthogonal AFDM. The input-output relationship reveals that non-orthogonality causes energy spreading across the effective channel matrix, creating a unique interference pattern dependent on $\alpha$ .

B. Inter-Carrier Interference (ICI) Analysis

The paper derives a closed-form expression for the ICI caused by non-orthogonal modulation.

The interference is characterized by a correlation matrix $C_\alpha$ .
The analysis shows that ICI magnitude is inversely proportional to $\alpha$ ; smaller $\alpha$ leads to stronger interference.
Crucially, the interference is not uniform; it exhibits a specific distribution where certain subcarrier pairs may retain orthogonality depending on the value of $\alpha$ and the subcarrier separation.

C. Detection Algorithm: Soft Iterative Detection (Soft ID)

To mitigate the severe ICI introduced by non-orthogonality, the authors propose a Soft Iterative Detection (Soft ID) algorithm:

Initial Estimation: Uses a Minimum Mean Square Error (MMSE) detector.
Iterative Cancellation: Utilizes symbol posterior probabilities (soft information) to estimate and subtract interference from other subcarriers in successive iterations.
Soft Clipping: Constrains soft symbol estimates to valid constellation regions to prevent error propagation.
Redetection: A specific step targets low-reliability symbols (identified by high estimation variance) for re-evaluation based on Mean Squared Error (MSE) criteria, refining the final decision.
Low-Complexity Pruning: To reduce computational load, a magnitude-based ICI pruning technique is introduced. It retains only the dominant $D$ interference terms (largest off-diagonal elements of $C_\alpha$ ) and treats the rest as noise, allowing a flexible trade-off between complexity and performance.

3. Key Contributions

Novel Waveform Design: Introduction of nAFDM, which achieves higher Spectral Efficiency than conventional AFDM by enabling controllable subcarrier overlapping via the bandwidth compression factor $\alpha$ .
Efficient Implementation: Development of an IDFT-based signal generation and demodulation scheme that allows practical implementation using standard IFFT/FFT modules without additional hardware complexity.
Theoretical Analysis: Derivation of the closed-form ICI expression, explicitly linking interference structure to the compression factor $\alpha$ and revealing the energy spreading behavior in doubly selective channels.
Advanced Detection: Proposal of a Soft ID algorithm with a redetection step and a low-complexity ICI pruning strategy, effectively managing the trade-off between Bit Error Rate (BER) and computational complexity.

4. Simulation Results

Extensive simulations were conducted under doubly selective channels (Jakes model) and Extended Vehicular A (EVA) channels:

BER Performance:
- nAFDM with Soft ID significantly outperforms SEFDM, OFDM, and OCDM.
- At a BER of $10^{-4}$ , nAFDM achieves an SNR gain of >10 dB over OFDM/OCDM and performs nearly identically to conventional orthogonal AFDM, despite the non-orthogonal nature.
- It outperforms uncoded AFDM by approximately 2.4–4.6 dB depending on $\alpha$ .
Spectral Efficiency:
- nAFDM achieves a 17.6% SE gain over OFDM and OCDM.
- SE increases as $\alpha$ decreases (e.g., $\alpha=0.85$ or $0.9$), offering a tunable trade-off between SE and BER.
Complexity vs. Performance:
- The ICI pruning technique (reducing $D$ ) allows for significant complexity reduction (e.g., 25–37.5%) with minimal BER loss.
- The proposed scheme maintains robustness even with imperfect Channel State Information (CSI).
Comparison with OTFS: nAFDM achieves BER and SE performance nearly identical to a non-orthogonal OTFS (nOTFS) variant but with lower modulation complexity.

5. Significance

This work addresses a critical bottleneck in 6G high-mobility communications: the need for spectral agility without sacrificing reliability.

Practicality: By leveraging existing FFT/IFFT hardware and proposing a low-complexity detection scheme, the paper bridges the gap between theoretical non-orthogonal gains and practical implementation.
Robustness: It demonstrates that non-orthogonal signaling can be made robust against high Doppler shifts, a feat previously thought difficult for non-orthogonal waveforms like SEFDM.
Future Impact: The nAFDM framework offers a promising candidate for next-generation wireless systems (6G), particularly for scenarios requiring high data rates in high-speed environments (e.g., HSR, satellite, UAVs), providing a flexible mechanism to balance spectral efficiency and error performance.