Delay-Doppler Domain Modulation for High-Mobility Communications

312 concepts242 papers1 proposal

In traditional wireless communications, data is typically modulated across a grid of time and frequency slots. This approach, known as Orthogonal Frequency Division Multiplexing (OFDM), serves as the backbone for 4G and 5G networks. OFDM works beautifully when the transmitter and receiver are stationary or moving slowly.

However, when deployed in high-mobility environments—such as high-speed trains traveling at 350 km/h, autonomous vehicles, Unmanned Aerial Vehicles (UAVs), or Low Earth Orbit (LEO) satellites—OFDM breaks down. The culprit is the Doppler effect. High relative velocities introduce significant frequency shifts that destroy the mathematical orthogonality between subcarriers. The subcarriers spill into one another, causing severe inter-carrier interference (ICI) and crippling the reliability of the wireless link.

To overcome this fundamental limitation, researchers are shifting away from the traditional time-frequency paradigm. Instead, they are modulating signals in the Delay-Doppler (DD) domain. By aligning the modulation domain with the underlying physical geometry of the wireless channel, this approach renders high-speed mobility an advantage rather than a destructive force.

The Breakdown of Orthogonality

To understand why delay-doppler modulation works, we must first look at why time-frequency systems fail at high speeds.

In an OFDM system, the available bandwidth is divided into closely spaced subcarriers. Orthogonality means that at the exact peak frequency of one subcarrier, the signal power of all other subcarriers is precisely zero. As Guangyao Liu et al. note in Pre-Chirp-Domain Index Modulation for Affine Frequency Division Multiplexing, high-speed mobility creates a "doubly dispersive" channel. The signal experiences both multi-path fading (dispersion in time due to different reflection paths) and Doppler shifts (dispersion in frequency due to movement).

Traditional OFDM (Time-Frequency Domain):
Frequency
   ▲
   │   ┌───┐ ┌───┐ ┌───┐
   │   │   │ │   │ │   │  ◄─── Subcarriers are perfectly spaced
   │   └───┘ └───┘ └───┘
   └────────────────────────► Time

With High-Mobility Doppler Shifts:
Frequency
   ▲
   │   ┌─┬─┐ ┌─┬─┐ ┌─┬─┐
   │   │ │ │ │ │ │ │ │ │  ◄─── Subcarriers shift and overlap!
   │   └─┴─┘ └─┴─┘ └─┴─┘       (Inter-Carrier Interference)
   └────────────────────────► Time

When a receiver moves rapidly, every reflection path from the surrounding environment arrives with a slightly different Doppler shift. This causes the subcarriers to shift unpredictably along the frequency axis, overlapping with their neighbors. The resulting inter-carrier interference acts as an inescapable noise floor. Furthermore, as An Le Ha et al. explain in Deep Learning-Aided 5G Channel Estimation, these rapid fluctuations mean the channel characteristics change faster than traditional estimation algorithms can track, leading to high bit error rates.

The Geometry of the Delay-Doppler Domain

Instead of trying to track a rapidly fluctuating time-frequency channel, delay-doppler modulation changes the coordinate system entirely.

Consider the physical environment between a transmitter and a mobile receiver. The wireless signal bounces off various reflectors (buildings, mountains, other vehicles). Each path can be characterized by two stable physical properties:

  1. Delay ($\tau$): The time it takes for the wave to travel the length of that specific path.
  2. Doppler shift ($\nu$): The frequency shift caused by the relative velocity of the vehicle along that path.

Crucially, while the phase of the radio wave fluctuates incredibly fast in the time-frequency domain—causing deep, unpredictable fades—the physical positions and velocities of the reflectors change very slowly.

Physical Environment             Delay-Doppler Domain Representation

   [Reflector 1]                    Doppler (Velocity)
       /   \                          ▲
      /     \                         │      • (Path 2: High speed, short delay)
     ▼       ▼                        │
[Tx] ───────► [Rx] (Moving Fast)      │  • (Path 1: Low speed, long delay)
     \       /                        │
      ▼     ▼                         └────────────────────────► Delay (Distance)
   [Reflector 2]                     (Channel appears as static, sparse points)

By mapping the channel to the Delay-Doppler domain, a highly volatile, time-varying channel is transformed into a quasi-static, sparse representation. A handful of isolated, stable points on a 2D grid capture the entire environment. This is the core intuition behind Orthogonal Time Frequency Space (OTFS) modulation and its variants.

Transforming the Signal

To exploit this stable geometry, information symbols must be modulated directly onto the 2D Delay-Doppler grid. The system then uses a series of mathematical transforms to convert this grid into a continuous time waveform for transmission, and reverses the process at the receiver.

In an OTFS system, the transformation pipeline typically works as follows:

  1. The Inverse Symplectic Finite Fourier Transform (ISFFT): This operation maps the data symbols from the 2D Delay-Doppler grid to the 2D Time-Frequency grid.
  2. The Heisenberg Transform: This converts the time-frequency grid into a 1D continuous-time signal that can be transmitted over the air.

Because of the ISFFT, every single data symbol placed on the Delay-Doppler grid is mathematically spread across the entire time-frequency block.

This localized mapping yields a massive advantage: full time-frequency diversity. In standard OFDM, if a single subcarrier hits a deep fade or experiences localized interference, the symbol on that subcarrier is lost. In an OTFS or delay-doppler system, because every symbol experiences the exact same integrated channel response across the entire block, the energy from a faded subcarrier is easily recovered from the rest of the block.

Alternative Formulations: AFDM and ODDM

To further optimize this domain mapping, researchers have introduced alternative waveforms that alter how the signal is structured:

  • Affine Frequency Division Multiplexing (AFDM): As explored by Guangyao Liu et al. in Pre-Chirp-Domain Index Modulation for Affine Frequency Division Multiplexing, AFDM utilizes orthogonal chirp signals (signals that change frequency over time) via the Discrete Affine Fourier Transform. This allows the system to perfectly adapt to the specific Doppler spread of the channel by tuning the chirp parameters, minimizing interference without increasing power consumption.
  • Orthogonal Delay-Doppler Multiplexing (ODDM): Proposed by Dezhi Wang et al. in Exploring Channel Estimation and Signal Detection for ODDM-based ISAC Systems, ODDM achieves a more precise orthogonal localization directly in the delay-Doppler domain. By avoiding the staggered boundaries of traditional OTFS transforms, ODDM reduces out-of-band emissions and provides even cleaner separation between overlapping paths.

Overcoming the Detection Challenge

While delay-doppler modulation simplifies the channel representation, the receiver still faces a challenge: separating the overlapping delay and Doppler paths to decode the original symbols. Because the received signal is a 2D convolution of the transmitted symbols and the channel response, standard linear equalization techniques can become computationally overwhelming.

To make real-time decoding feasible, recent innovations leverage structural and learning-based approaches. For instance, Jiarui Xu et al. introduce a specialized neural network framework in 2D-RC: Two-Dimensional Neural Network Approach for OTFS Symbol Detection. By embedding domain-specific knowledge—specifically the 2D circular convolution properties of the delay-Doppler channel—directly into a Reservoir Computing architecture, the authors achieve near-optimal symbol detection with a fraction of the computational complexity required by traditional model-based iterative detectors.

Applications in Next-Generation Networks

The inherent immunity of delay-doppler waveforms to Doppler-induced orthogonality loss makes them a foundational technology for 6G architectures, particularly in two expanding domains:

Integrated Sensing and Communication (ISAC)

Radar systems naturally operate in the delay-Doppler domain; they calculate an object's distance via time delay and its speed via the Doppler shift. By using a delay-doppler communication waveform like OTFS or ODDM, wireless networks can perform dual functions simultaneously.

As Weijie Yuan et al. demonstrate in From OTFS to DD-ISAC: Integrating Sensing and Communications in the Delay Doppler Domain, a single transmitted signal can carry high-throughput data to a mobile user while its reflections are parsed by the base station to track surrounding objects with radar-like precision. This native domain alignment eliminates the need to split time or frequency resources between sensing and communication.

Satellite and Aerial IoT Networks

In LEO satellite systems, the extreme velocities of satellites relative to ground terminals create massive Doppler shifts that terrestrial protocols cannot handle. In Active Terminal Identification, Channel Estimation, and Signal Detection for Grant-Free NOMA-OTFS in LEO Satellite Internet-of-Things, Xingyu Zhou 0009 et al. show that combining OTFS with Non-Orthogonal Multiple Access (NOMA) allows satellites to reliably identify and decode signals from thousands of low-power IoT devices simultaneously, bypassing the synchronization failures that plague traditional satellite uplinks.

By shifting the mathematical foundation of wireless signaling from volatile time-frequency coordinates to the stable physical geometry of delay and Doppler, modern communication systems can finally maintain high-capacity, reliable links at extreme velocities.

History & citationVersion 2 · 2026-10-04

Current · Permanent link

  1. Version 2 (latest) · 2026-10-04 · 236 papers
  2. Version 1 · 2026-10-04 · 180 papers
Cite this direction

Latest version (a living page):

@misc{lacuna_inter_carrier_interference_2c7112e3,
  title = {inter-carrier interference},
  author = {{Lacuna}},
  howpublished = {Lacuna research direction},
  year = {2026},
  url = {https://lacuna.tiptreesystems.com/direction/inter-carrier-interference/rdr_2c7112e31a9f436b846bf1e2be71c814},
  note = {Latest version of a living research direction; cite a version for a fixed text. Accessed 2026-11-19},
}

This exact version (version 2, frozen text):

@misc{lacuna_inter_carrier_interference_2c7112e3_v2,
  title = {inter-carrier interference},
  author = {{Lacuna}},
  howpublished = {Lacuna research direction},
  year = {2026},
  url = {https://lacuna.tiptreesystems.com/direction/inter-carrier-interference/rdr_2c7112e31a9f436b846bf1e2be71c814/v/2},
  note = {Version 2, frozen 2026-10-04. Accessed 2026-11-19},
}
About

These directions overlap with this one through shared papers.

See something wrong? Report a mistake