In traditional wireless communications, multi-path propagation is a persistent challenge. When a transmitter sends a signal, the radio waves bounce off buildings, mountains, and other obstacles, creating multiple "echoes" that arrive at the receiver at different times. This time dispersion causes inter-symbol interference (ISI), where trailing echoes of previous symbols corrupt the arrival of the current symbol.
To combat ISI, modern wireless standards like 5G rely heavily on Orthogonal Frequency-Division Multiplexing (OFDM). OFDM splits a high-rate data stream into many low-rate subcarriers, adding a "cyclic prefix" guard interval to absorb the echoes. However, this approach comes with significant costs: high power overhead, a high peak-to-average power ratio (PAPR), and complex receiver-side equalization hardware.
A compelling alternative is Delay Alignment Modulation (DAM). Instead of trying to clean up or tolerate multi-path echoes at the receiver, DAM uses the massive spatial degrees of freedom provided by large antenna arrays (Massive MIMO) to pre-compensate for these delays at the transmitter. By intentionally delaying signals before they are even transmitted, DAM transforms a hostile, time-dispersive channel into a simple, interference-free one.
The Core Intuition: Flipping the Delay Spread
To understand why DAM works, imagine a multi-path channel where a transmitter's signal reaches a receiver via three distinct paths: a direct path (short delay), a medium-length path (medium delay), and a long path (long delay). In a standard single-carrier transmission, a single symbol sent at time $t$ will hit the receiver three separate times, blurring into subsequent symbols.
DAM flips this dynamic entirely through spatial-delay pre-compensation. As described by Yong Zeng 0001, the transmitter uses its large antenna array to isolate each spatial path via beamforming, and then deliberately introduces a specific time offset to the symbols sent along each path.
If the longest path takes an extra 10 microseconds to arrive compared to the shortest path, the transmitter will intentionally delay the signal sent along the shortest path by exactly 10 microseconds. By customizing these intentional delays to perfectly counteract the environmental propagation delays, the transmitter ensures that all multi-path components arrive at the receiver simultaneously and constructively.
Traditional Multi-Path:
Transmitter ----(Path 1: Short Delay)----> ┐
----(Path 2: Long Delay)-----> ┴─> Arrive at different times (ISI)
Delay Alignment Modulation:
Transmitter ----(Intentional Delay + Path 1)----> ┐
----(No Added Delay + Path 2)-------> ┴─> Arrive simultaneously (Constructive Boost)
As Yong Zeng 0001 demonstrates, this mechanism effectively shrinks or entirely eliminates the channel's delay spread from the perspective of the receiver. What was once a frequency-selective channel plagued by interference becomes an equivalent flat-fading or additive white Gaussian noise (AWGN) channel. Environmental echoes are no longer a source of distortion; instead, they act as a constructive signal boost, significantly increasing the received signal-to-noise ratio (SNR) without requiring complex receiver-side equalization.
Scaling to Multi-User and Multi-IRS Networks
The benefits of DAM become even more pronounced when scaled to complex multi-user and multi-surface environments. In millimeter-wave (mmWave) massive MIMO networks, channels are naturally sparse, meaning signals travel along a few well-defined paths. Xingwei Wang 0013 et al. showed that by combining pre-emptive temporal alignment with spatial steering, a base station can eliminate both ISI and inter-user interference (IUI) simultaneously. Because the spatial resolution of mmWave arrays is so high, path-based beamforming can direct aligned signals to specific users without bleeding into others (Xingwei Wang et al.).
This framework can also be extended to environments assisted by Intelligent Reflecting Surfaces (IRS)—smart mirrors that redirect radio waves. Yong Zeng 0001 et al. introduced single-carrier DAM for multi-IRS systems, demonstrating that joint spatial and temporal processing can align paths bouncing off multiple distributed surfaces. In a similar vein, Rongen Dong et al. utilized distributed active IRS units mounted on UAVs, combining transmitter null-space projection with reflector phase alignment to enable high-speed, multi-stream transmission free from inter-stream interference.
Advanced Alignments: Double-Side and Delay-Doppler
While compensating for delays entirely at the transmitter is elegant, it requires highly precise transmitter-side channel state information and can occasionally demand high peak transmission power. To optimize this trade-off, Xingwei Wang et al. proposed double-side delay alignment modulation. By strategically splitting the delay compensation burden between both the transmitter and the receiver, this dual-sided approach eliminates ISI while preserving valuable signal power, leading to superior spectral efficiency in terahertz and mmWave communications.
Furthermore, real-world wireless channels are not just time-dispersive; they are also time-varying. When transmitters or receivers move, the signals experience frequency shifts known as Doppler spreads. To address this, Yong Zeng 0001 developed Delay-Doppler Alignment Modulation (DDAM).
DDAM applies both specific time delays and frequency shifts to the signals for each propagation path at the transmitter. This ensures that the components arrive at the receiver synchronized in both time and frequency domains. By shifting the burden of both delay and Doppler equalization to the transmitter, DDAM transforms a highly volatile, doubly selective channel into a stable, time-invariant one.
The Broader Paradigm of Waveform Superposition
The underlying philosophy of DAM—manipulating wireless signals so they combine constructively in the air—is part of a broader shift toward treating wireless superposition as a tool rather than a nuisance.
This concept is foundational to Over-the-Air Computation (AirComp), an architectural paradigm for massive IoT data aggregation explored by Guangxu Zhu et al.. Instead of preventing devices from interfering with each other, AirComp requires devices to perform magnitude and phase alignment based on local channel states so that their signals combine in the air to directly compute an aggregate value (like an average).
A practical application of this is found in federated learning, where a central server needs to aggregate local model updates from thousands of IoT devices. Huayan Guo et al. developed a two-stage pre-equalization protocol to compensate for timing and frequency offsets across devices, proving that accurate over-the-air waveform superposition can drastically accelerate distributed learning. Even in channel estimation, Xiliang Luo et al. demonstrated that intentionally varying the alignment patterns of overlapping signals across time slots allows a system to cleanly resolve and estimate individual multi-path components that would otherwise be indistinguishable.
By shifting the complexity of multi-path management from receiver-side mitigation to transmitter-side alignment, Delay Alignment Modulation and its extensions offer a scalable, power-efficient path forward for next-generation wireless networks.
Key Foundational Work
Over-the-Air Computing for Wireless Data Aggregation in Massive IoT (2020) — contributed 1 concept elements to this direction.
A Scalable Framework for CSI Feedback in FDD Massive MIMO via DL Path Aligning (2017) — contributed 1 concept elements to this direction.