Balancing Throughput and Detection Risk in Covert Communications

24 concepts18 papers1 proposal

In wireless security, cryptography ensures secrecy—preventing an adversary from decoding a message. However, in many high-stakes scenarios, the goal is covertness—preventing an adversary (often called a "warden") from even realizing a transmission is taking place.

As Yingxin Zhang et al. demonstrate, transitioning from standard secrecy to full covertness requires a significant sacrifice in transmission speed. To remain hidden, a signal must be statistically indistinguishable from background noise or environmental interference. This reality introduces a fundamental physical bottleneck: increasing transmit power improves the signal-to-noise ratio (SNR) and boosts throughput, but it simultaneously pushes the signal above the noise floor, exposing it to the warden.

According to Hao Jiang et al. and Han-Bae Kong et al., tightening the covertness constraint directly chokes the maximum achievable data rate. This trade-off spans the entire network stack, from low-level routing up to application-layer services. For instance, Ruichen Zhang et al. highlight that hiding sensitive Large Language Model (LLM) queries within wireless traffic requires dropping transmit power to a level where transmission errors and latency spike significantly.

To overcome this bottleneck, recent research focuses on expanding the covertness-throughput envelope by exploiting the warden's detection mechanics, engineering environmental uncertainty, and leveraging spatial degrees of freedom.


Exploiting the Warden's Detection Mechanics

To maximize throughput under a strict covertness constraint, a transmitter must exploit how a warden processes signals. Traditional security models assume the warden performs batch processing over a fixed window. However, real-world adversaries monitor channels continuously.

Ke-Wen Huang et al. reframe this dynamic by modeling the warden using Sequential Change-Point Detection (SCPD). In a sequential test, the warden continuously samples the channel to detect a statistical shift from a "noise-only" state to a "signal-plus-noise" state. Because statistical tests inherently require a finite number of samples to accumulate sufficient confidence, there is an inevitable detection delay.

Transmission Start
       │
       ▼
───────┼──────────────────────────────► Time
       │◄─────── Detection Delay ──────►│
                                       ▼
                                 Warden Triggers

The authors demonstrate that a transmitter can exploit this temporal window. By jointly optimizing transmit power and burst duration, the system can transfer a high-rate burst of data and cease transmission just before the warden's sequential detector accumulates enough evidence to trigger an alarm.


Engineering Environmental Uncertainty

If a covert signal must blend into the background, a logical countermeasure is to manipulate the background itself. Wardens rely on a stable, predictable noise baseline to detect anomalies. By injecting uncertainty into this baseline, transmitters can safely operate at higher power levels.

Cooperative Jamming

One method to disrupt the warden's baseline is through intentional interference. Tong-Xing Zheng et al. propose a distributed, uncoordinated jamming strategy where multiple friendly auxiliary nodes act as jammers. By activating only when their localized interference to the legitimate receiver is low, these nodes artificially inflate and vary the noise floor at the warden. This dynamic masking allows the primary transmitter to scale up its power—and its throughput—without crossing the warden's detection threshold.

Ambient Interference

Instead of generating artificial noise, a system can hide within existing network traffic. Shaohan Feng et al. analyze device-to-device (D2D) communications underlaying a larger cellular network. Here, the aggregate, stochastic interference generated by standard cellular users acts as a natural camouflage. The authors show that by treating cellular interference as a protective shroud, D2D nodes can balance their power levels to maintain link reliability while remaining invisible to external adversaries.

Temporal and Probabilistic Uncertainty

When a device relies on energy harvesting, its transmission patterns can become highly predictable, allowing a warden to establish a clean baseline during periods when the device is known to be charging. To counter this, Yida Wang 0004 et al. introduce a probabilistic "accumulate-then-transmit" protocol.

Instead of greedily transmitting the moment enough energy is harvested, the device intentionally skips transmission opportunities based on a calculated probability. This introduces strategic silence, denying the warden a predictable operational baseline and maximizing the statistical ambiguity of the channel.


Spatial Degrees of Freedom: Beamforming and Intelligent Surfaces

The ideal covert communication system would deliver high power to the legitimate receiver while ensuring the signal drops completely into the noise floor at the warden's coordinates. Achieving this requires precise control over the spatial distribution of electromagnetic waves.

Reconfigurable Intelligent Surfaces (RIS)

RIS and Simultaneously Transmitting and Reflecting RIS (STAR-RIS) platforms provide a way to dynamically reshape the wireless environment. Hai Yang et al. and Xinying Chen et al. demonstrate that by jointly optimizing transmission power and the phase shifts of an intelligent surface, the system can decouple the legitimate channel from the warden's channel.

The surface elements are configured to reflect signals destructively toward the warden—keeping the received power below the noise floor—while aligning them constructively at the intended recipient to maximize data rates.

                  ┌───────────────┐
                  │   STAR-RIS    │
                  └───────┬───────┘
                         ╱ ╲
  Constructive          ╱   ╲          Destructive
  Alignment            ╱     ╲         Cancellation
                      ▼       ▼
             ┌───────────┐ ┌───────────┐
             │ Receiver  │ │  Warden   │
             │ (High SNR)│ │ (In Noise)│
             └───────────┘ └───────────┘

Advanced Antenna Architectures

Beyond digital beamforming, hardware-level adaptations offer new ways to suppress signals at adversary locations:

  • Rotatable Antennas (RAs): Qi Dai et al. explore the use of antennas that physically adjust their geometric orientation (boresight). This mechanical degree of freedom, combined with digital beamforming, allows the transmitter to carve out deep physical nulls in the direction of distributed adversaries, even when those adversaries overlap with the main transmission path.
  • Pinching-Antenna Systems (PASS): Hao Jiang et al. utilize flexible antenna placements to boost covert rates, demonstrating that spatial flexibility can overcome the geometric advantages often held by well-positioned eavesdroppers.
  • Hardware Imperfections: Yulin Teng et al. demonstrate that inherent hardware characteristics, such as mutual coupling in millimeter-wave antenna arrays, can be repurposed. They leverage these unique physical fingerprints to perform secure user authentication while simultaneously maintaining a covert channel during the initial beam-alignment phase.

Multi-Functional Conflicts: Sensing vs. Stealth

The trade-off between throughput and covertness becomes more complex when wireless systems are tasked with objectives beyond data transfer, such as in Integrated Sensing and Communications (ISAC) frameworks.

As detailed by Hongyang Du et al. and Runzhe Tang et al., a fundamental physical conflict exists between radar sensing and covert communication. Effective radar sensing requires highly structured, high-power waveforms to travel long distances, reflect off targets, and return with sufficient fidelity for accurate tracking.

This requirement directly contradicts the low-power, noise-like, unstructured waveforms needed for covertness. High-fidelity sensing requirements constrain how a base station can shape its artificial noise and allocate its power, shrinking the optimization space available for hiding data transmissions.

┌─────────────────────────────────────────────────────────────────┐
│                    Shared Spectrum Waveform                     │
└────────────────────────────────┬────────────────────────────────┘
                                 │
                ┌────────────────┴────────────────┐
                ▼                                 ▼
   ┌───────────────────────────┐     ┌───────────────────────────┐
   │       Radar Sensing       │     │   Covert Communication    │
   ├───────────────────────────┤     ├───────────────────────────┤
   │ • High Power              │  ◄──┼──► • Low Power            │
   │ • Structured Waveforms    │     │ • Noise-like Waveforms    │
   │ • Long Range              │     │ • Low Profile             │
   └───────────────────────────┘     └───────────────────────────┘

A similar spatial tension occurs in mobile surveillance. Hailong Huang et al. examine a team of Unmanned Aerial Vehicles (UAVs) collaboratively eavesdropping on ground transmitters. To maximize the SNR of the intercepted data, the UAVs need to fly as close to the targets as possible. However, closer proximity increases their visibility, compromising their visual disguise.

Whether managing the power of an ISAC waveform or the physical trajectory of a surveillance drone, modern covert systems must balance operational performance against the risk of detection.

Key Foundational Work

Probabilistic Accumulate-then-Transmit in Wireless-Powered Covert Communications (2021) — contributed 3 concept elements to this direction.
Wireless Covert Communications Aided by Distributed Cooperative Jamming over Slow Fading Channels (2021) — contributed 2 concept elements to this direction.
Navigation of a UAV Team for Collaborative Eavesdropping on Multiple Ground Transmitters (2021) — 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_covertness_throughput_trade_off_63caa9c7,
  title = {covertness-throughput trade-off},
  author = {{Lacuna}},
  howpublished = {Lacuna research direction},
  year = {2026},
  url = {https://lacuna.tiptreesystems.com/direction/covertness-throughput-trade-off/rdr_63caa9c7a0424af88aefa7af93cdec00},
  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_covertness_throughput_trade_off_63caa9c7_v1,
  title = {covertness-throughput trade-off},
  author = {{Lacuna}},
  howpublished = {Lacuna research direction},
  year = {2026},
  url = {https://lacuna.tiptreesystems.com/direction/covertness-throughput-trade-off/rdr_63caa9c7a0424af88aefa7af93cdec00/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