Analog Spatial Filtering to Overcome Quantization Bottlenecks

33 concepts13 papers1 proposal

In modern wireless communications, receivers are increasingly caught in a hardware vice. To capture wide bandwidths or process signals from massive antenna arrays, hardware designers must use analog-to-digital converters (ADCs) with high sampling rates or low power profiles. However, an unyielding physical trade-off governs converter design: as an ADC's sampling rate increases or its power budget decreases, its resolution (bit depth) drops.

This creates a severe vulnerability to high-power interference, such as malicious jamming or a full-duplex system's own transmitted signal. When a massive interferer hits a low-resolution ADC, it creates an information bottleneck that cannot be resolved by digital software alone. To overcome this, a growing body of research focuses on analog-domain spatial filtering—manipulating the radio frequency (RF) signal before it reaches the ADC to shield the hardware's limited dynamic range.

The Irreversible Wall of Quantization Noise

To understand why analog-domain intervention is necessary, we must first look at how a low-resolution ADC behaves in the presence of strong interference. As quantified by Laska et al. in Beyond Nyquist: efficient sampling of sparse bandlimited signals and The Pros and Cons of Compressive Sensing for Wideband Signal Acquisition: Noise Folding versus Dynamic Range, every missing bit of ADC resolution reduces the receiver's dynamic range by approximately 6 dB. When a receiver encounters a high-power interferer, it faces a lose-lose scenario:

  1. Clipping: If the receiver's gain is set too high, the large interferer exceeds the maximum voltage of the ADC, causing severe signal clipping and irreversible distortion, as noted by Azar et al. in Residual Recovery Algorithm for Modulo Sampling.
  2. Quantization Noise Flooding: To avoid clipping, the receiver's Automatic Gain Control (AGC) must scale down the entire incoming signal so that the large interferer fits within the ADC's voltage range. However, because the ADC has very few bits, this scaling forces the weak, legitimate signal down into the quantization noise floor.

As Marti et al. demonstrate in Hybrid Jammer Mitigation for All-Digital mmWave Massive MU-MIMO, once a weak signal is buried beneath this expanded quantization noise floor, the information loss is mathematically absolute. This is a universal hardware constraint: if the input signal's scaling is mismatched to the converter's dynamic range, the system fails to utilize its bit-depth effectively, a phenomenon observed even in specialized high-precision sensing systems by Yao in Research on High Precision Temperature Acquisition System for Radiofrequency Ablation System.

Why Digital Mitigation Fails

A common misconception is that multi-antenna systems (MIMO) can easily bypass this issue by using digital beamforming to "null" the interference. In theory, an array with dozens of antennas can combine signals digitally to cancel out a spatial jammer.

However, this assumption breaks down under finite-resolution quantization. In Fundamental Limits for Jammer-Resilient Communication in Finite-Resolution MIMO, the authors prove that finite-resolution ADCs place a hard ceiling on interference rejection that cannot be overcome by simply adding more antennas. They derive a fundamental scaling law: to maintain reliable communication, a receiver requires exactly one additional bit of ADC resolution for every 6.02 dB increase in jamming power. Without these bits, the quantization process destroys the precise spatial phase information required to perform digital nulling, causing the achievable communication rate to collapse to zero.

[Incoming RF Signals] ──> [AGC Scales Down] ──> [Low-Res ADC] ──> [Digital Domain]
  (Legitimate Signal +                              │                 │
   Massive Jammer)                                  ▼                 ▼
                                            Quantization Noise   Digital Nulling Fails
                                            Buries Weak Signal   (Information Already Lost)

This bottleneck is vividly illustrated in In-Band Full-Duplex (IBFD) wireless systems, where a device transmits and receives on the same frequency simultaneously. As surveyed by Sabharwal et al. in In-Band Full-Duplex Wireless: Challenges and Opportunities, the deafening roar of self-interference from the local transmitter easily saturates the receiver's ADC. Roberts et al. show in Hybrid Beamforming for Millimeter Wave Full-Duplex Under Limited Receive Dynamic Range that if this self-interference is not suppressed before the ADC, the resulting quantization noise floor completely negates the spectral efficiency gains of full-duplex operation, making it perform worse than traditional half-duplex systems.

Because digital post-processing cannot recover what is lost at the quantization interface—a point reinforced by Shlezinger et al. in Asymptotic Task-Based Quantization With Application to Massive MIMO—the interference must be managed in the analog domain.

Shifting the Burden to Analog Spatial Transforms

The core mechanism of analog-domain spatial filtering is to exploit the spatial separation between the legitimate users and the interferer before the signals are digitized. By applying an analog linear transform (using networks of phase shifters, attenuators, or directional couplers), the receiver can alter how signal energy is distributed across the ADC bank.

In a straightforward implementation, the analog front-end can be configured to place a spatial null directly in the direction of the interference. As shown by Zhang et al. in Analog-Domain Suppression of Strong Interference Using Hybrid Antenna Array, performing this nulling in the analog domain can suppress strong interference by over 65 dB. This drastically relaxes the dynamic range requirements for the subsequent ADCs, allowing the system to use low-resolution, low-power converters without risking saturation or noise-floor flooding.

[Incoming RF Signals] ──> [Analog Spatial Filter] ──> [Low-Res ADCs] ──> [Digital Domain]
  (Legitimate Signal +         (Nulls or Isolates       (Protected from     (Clean Signal
   Massive Jammer)               Interference)            Saturation)        Recovered)

Spatial Isolation: Beam-Slicing and Householder Reflections

Perfect analog nulling requires highly precise, expensive, and adaptive analog hardware to track the exact angle of an interferer. To make this approach practical for low-cost millimeter-wave (mmWave) massive MIMO arrays, researchers have developed hybrid architectures that favor spatial isolation over perfect cancellation.

Instead of trying to erase the jammer entirely in analog, these architectures use fixed or coarse analog transforms to isolate the jammer's energy into a tiny subset of ADCs, leaving the remaining ADCs uncorrupted.

Beam-Slicing (SNIPS)

Proposed by Castañeda et al. in Beam-Slicing for Jammer Mitigation in mmWave Massive MU-MIMO and expanded by Marti et al. in Jammer Mitigation via Beam-Slicing for Low-Resolution mmWave Massive MU-MIMO, this technique applies a localized spatial transform that acts like a prism. It concentrates the high-power jammer energy into just one or two "slices" (ADCs).

While those specific ADCs are completely blinded by quantization noise or clipping, the remaining ADCs in the system receive zero jammer energy. Legitimate user signals, which arrive from different spatial angles, are distributed across all the ADCs. The digital baseband can then simply discard or "soft-null" the corrupted channels and still perfectly reconstruct the user data from the remaining clean channels.

Householder Reflections (HDR MIMO)

A mathematically elegant variation of this concept is introduced by Palhares et al. in High Dynamic Range mmWave Massive MU-MIMO with Householder Reflections. In scenarios where users have vastly different signal strengths (the near-far problem), a dominant nearby user can act as an unintended interferer to a weak, distant user.

The authors propose an analog front-end that implements an adaptive Householder reflection matrix. This orthogonal transform mathematically rotates the incoming signal vector so that the entire energy of the dominant signal is mapped onto a single ADC. By isolating the high-power signal to a single converter, the AGCs for all the other ADCs can remain highly sensitive, ensuring that the weak users are captured with high fidelity and are not drowned out by quantization noise.

By combining the power savings of low-resolution hardware with the protective shielding of analog spatial preprocessing, these hybrid architectures successfully bypass the fundamental limits of digital-only receivers, paving the way for interference-resilient, high-bandwidth wireless networks.

Key Foundational Work

The Pros and Cons of Compressive Sensing for Wideband Signal Acquisition: Noise Folding versus Dynamic Range (2012) — contributed 3 concept elements to this direction.
Beyond Nyquist: Efficient Sampling of Sparse Bandlimited Signals (2010) — contributed 2 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_analog_domain_spatial_filtering_460f8e4d,
  title = {analog-domain spatial filtering},
  author = {{Lacuna}},
  howpublished = {Lacuna research direction},
  year = {2026},
  url = {https://lacuna.tiptreesystems.com/direction/analog-domain-spatial-filtering/rdr_460f8e4d8abc41a4805ab205ea054f6d},
  note = {Latest version of a living research direction; cite a version for a fixed text. Accessed 2026-11-21},
}

This exact version (version 1, frozen text):

@misc{lacuna_analog_domain_spatial_filtering_460f8e4d_v1,
  title = {analog-domain spatial filtering},
  author = {{Lacuna}},
  howpublished = {Lacuna research direction},
  year = {2026},
  url = {https://lacuna.tiptreesystems.com/direction/analog-domain-spatial-filtering/rdr_460f8e4d8abc41a4805ab205ea054f6d/v/1},
  note = {Version 1, frozen 2026-10-04. Accessed 2026-11-21},
}
About

These directions overlap with this one through shared papers.

See something wrong? Report a mistake