Distortion Limited Performance in Coarsely Quantized Systems

19 concepts12 papers1 proposal

In traditional wireless communication design, the primary obstacle to high data rates and reliable transmission is environmental noise—such as thermal noise or channel interference. The classic remedy is straightforward: increase the transmission power. By boosting the signal-to-noise ratio (SNR), the receiver can more easily distinguish the intended signal from background fluctuations.

However, this foundational intuition breaks down entirely when the communication chain contains low-resolution hardware or capacity-constrained digital links. When components like 1-bit Analog-to-Digital Converters (ADCs) or Digital-to-Analog Converters (DACs) are introduced to save power and cost, systems encounter a hard performance ceiling known as the quantization noise floor. In this high-SNR regime, turning up the transmission power yields diminishing returns and eventually fails to improve performance altogether.

Why Power Cannot Overcome Coarse Quantization

To understand why increasing power fails, consider the geometric effect of a 1-bit quantizer. A 1-bit ADC simply records the sign of an incoming signal, mapping any positive voltage to $+1$ and any negative voltage to $-1$.

If a transmitter increases its signal amplitude tenfold to overcome thermal noise, the input to the 1-bit receiver scales accordingly. However, the output of the quantizer remains exactly the same sequence of $+1$s and $-1$s. Beyond the threshold where the signal becomes stronger than the thermal noise, any additional transmit power scales the quantization distortion exactly in proportion to the signal itself.

As shown by Sven Jacobsson et al. in the context of massive MU-MIMO-OFDM downlinks, systems utilizing ultra-low-resolution converters reach a performance saturation point. Once this ceiling is hit, the internal distortion caused by coarse quantization dominates over external noise, rendering further power increases useless.

Traditional Noise-Limited Regime:
[Signal Power ↑] / [Fixed Thermal Noise] = SNR ↑ (Performance Improves)

Distortion-Limited Regime:
[Signal Power ↑] / [Quantization Distortion ↑] = SQNR Constant (Performance Plateaus)

This phenomenon is explicitly quantified by Dan Li et al. for Terahertz communications. The authors demonstrate that with 1-bit precoding, the achievable data rate eventually becomes completely independent of transmit power. Because the quantization noise floor scales alongside the signal, the signal-to-noise-and-distortion ratio (SNDR) flattens into a permanent plateau.

The Limits of Signal Processing and Noise Shaping

A natural countermeasure to this bottleneck is to use advanced signal processing or spatial noise-shaping to push quantization noise out of the signal band. For instance, spatial sigma-delta ($\Sigma\Delta$) architectures use antenna arrays to shift quantization errors into higher spatial frequencies.

Yet, even these sophisticated architectures cannot entirely bypass the information loss inherent to low-resolution hardware. In their work on millimeter-wave channel estimation, R. S. Prasobh Sankar et al. demonstrate in Millimeter Wave MIMO Channel Estimation with 1-bit Spatial Sigma-delta Analog-to-Digital Converters that while noise-shaping matches infinite-resolution hardware at low SNRs, a persistent performance gap emerges at high SNRs. In a follow-up study, the authors confirm that the quantization noise floor cannot be further suppressed by the noise-shaping mechanism once a certain signal power threshold is reached, creating an unyielding performance ceiling.

To mitigate these effects, precoding strategies must be redesigned from scratch to treat quantization distortion as a structural constraint rather than random noise. Yavuz Yapici et al. show that by explicitly accounting for non-linear 1-bit distortion within a Signal-to-Leakage-plus-Noise Ratio (SLNR) framework, precoders can better navigate the distortion floor and achieve significantly higher spectral efficiencies than traditional linear methods.

The Geometry of Error and Zero Diversity

The impact of the quantization floor becomes catastrophic when the hardware resolution drops below the requirements of the modulation scheme. If a system attempts to transmit a complex constellation (like 16-QAM) through a quantizer with fewer representation levels than the modulation order ($n < \log_2 M$), the receiver loses the geometric ability to distinguish between adjacent symbols.

Samiru Gayan et al. mapped this behavior in phase-modulated systems, revealing a "ternary behavior" where the decay exponent of the symbol error probability drops sharply if the number of quantization bits is insufficient. When $n < \log_2 M$, an irreducible error floor emerges that cannot be mitigated by any amount of signal amplification.

From an information-theoretic perspective, this mismatch results in a complete failure of channel diversity. Zheyu Wu et al. analyzed this breakdown in Diversity Order Analysis for Quantized Constant Envelope Transmission, proving that an insufficient quantization resolution forces the system into a state of zero diversity order. Because the signal is inherently mapped to incorrect decision boundaries by the hardware itself, the symbol error probability hits a flat, constant floor.

The Ubiquity of the Quantization Ceiling

The distortion-limited bottleneck is not merely a hardware issue at the antenna front-end; it is a fundamental property of any communication framework constrained by finite-capacity digital representations.

Oblivious Relays and Limited Backhaul

In distributed MIMO and multicell processing, base stations frequently act as "oblivious" relays that simply compress and forward received signals over a finite-capacity backhaul network to a central processor. H. Vincent Poor et al. and Osvaldo Simeone et al. demonstrate that while this oblivious strategy is highly effective at moderate data rates, it hits a strict performance ceiling at high SNRs. The quantization noise introduced by compressing the signals for the backhaul link becomes the dominant bottleneck, preventing the distributed system from scaling to its theoretical capacity limits.

Source-Channel Coding and Semantic Communications

The presence of a quantization floor also breaks classical communication theorems. For instance, the long-standing rule that "linear transmission is optimal" when sending Gaussian sources over Gaussian channels fails under quantization constraints. Morteza Varasteh et al. demonstrate that when a receiver uses low-resolution ADCs, non-linear encoding mappings are strictly required to minimize mean squared error and manage the distortion floor.

This saturation behavior extends to modern, deep-learning-based semantic communication systems. Shuai Ma 0002 et al. modeled this behavior using an empirical Alpha-Beta-Gamma formula, proving that the end-to-end performance plateau of semantic models is strictly bounded by the number of quantization bits allocated during source encoding.

Escaping the Ceiling via Analog Mapping

To completely bypass the quantization ceiling, some modern architectures abandon digital compression entirely in favor of direct analog mappings. Takuya Fujihashi et al. developed HoloCast, a transmission scheme for 3D point clouds that maps graph-transformed signals directly to analog waveforms. By avoiding the rigid boundaries of digital quantization, the system eliminates the irreducible distortion floor, allowing visual reconstruction quality to scale continuously and gracefully with the true strength of the wireless channel.

Key Foundational Work

Downlink Multicell Processing with Limited-Backhaul Capacity (2009) — contributed 2 concept elements to this direction.
Distributed MIMO Systems with Oblivious Antennas (2008) — contributed 1 concept elements to this direction.
History & citationVersion 1 · 2026-10-04

Current · Permanent link

View frozen version 1

Cite this direction

Latest version (a living page):

@misc{lacuna_distortion_limited_communication_d04772dc,
  title = {distortion-limited communication},
  author = {{Lacuna}},
  howpublished = {Lacuna research direction},
  year = {2026},
  url = {https://lacuna.tiptreesystems.com/direction/distortion-limited-communication/rdr_d04772dc24274ad6a7d3ccf6c7ad8e2f},
  note = {Latest version of a living research direction; cite a version for a fixed text. Accessed 2026-11-19},
}

This exact version (version 1, frozen text):

@misc{lacuna_distortion_limited_communication_d04772dc_v1,
  title = {distortion-limited communication},
  author = {{Lacuna}},
  howpublished = {Lacuna research direction},
  year = {2026},
  url = {https://lacuna.tiptreesystems.com/direction/distortion-limited-communication/rdr_d04772dc24274ad6a7d3ccf6c7ad8e2f/v/1},
  note = {Version 1, frozen 2026-10-04. Accessed 2026-11-19},
}
About

These directions overlap with this one through shared papers.

See something wrong? Report a mistake