⚡ electrical engineering

Covert Communication with Spatially Heterogeneous User Cooperation Against a Geometry-Aware Warden

This paper proposes a low-complexity framework for covert communication where a single user is assisted by spatially heterogeneous non-covert users against a geometry-aware warden, deriving closed-form detection metrics and optimizing power control to minimize the required number of cooperative users while balancing interference effects.

Original authors: Hyeonsik Yeom, Jinyoung Lee

Published 2026-08-12

📖 6 min read🧠 Deep dive

CC BY 4.0

Original authors: Hyeonsik Yeom, Jinyoung Lee

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine you are trying to send a secret note to a friend across a crowded, noisy room. You don't want the room's security guard, let's call him "Willie," to even know a note is being passed. If Willie hears a whisper, he might suspect something is up. This is the world of covert communication: a branch of wireless security where the goal isn't just to hide the message (like using a secret code), but to hide the very existence of the conversation.

To pull this off, you need to make your whisper sound exactly like the background noise. In the real world, wireless signals are messy. Sometimes you have interference from other devices, which can actually help hide your secret signal. But here's the tricky part: if you just turn up the volume on the background noise to cover your whisper, your friend might not hear you at all. You need a perfect balance. You need just enough "noise" to confuse Willie, but not so much that it drowns out your friend. This paper tackles the problem of how to coordinate a group of people to create that perfect, confusing noise without ruining the conversation for your friend.

The Big Idea: A Smart Crowd vs. A Geometry-Savvy Guard

In this study, researchers Hyeonsik Yeom and Jinyoung Lee set up a scenario with a secret sender (Alice), a receiver (Bob), a security guard (Willie), and a large crowd of regular people (the cooperative users). Alice wants to send a secret message to Bob. Willie is watching, trying to detect if Alice is talking. The crowd can help by shouting (transmitting interference) to mask Alice's voice.

The twist in this paper is that the researchers realized previous methods were too simple. They treated the crowd like a uniform blob of noise, assuming everyone was equally far from the guard and equally loud. But in reality, people are scattered everywhere. Some are close to Willie, some are far; some are loud, some are quiet. The guard, Willie, is "geometry-aware," meaning he knows the layout of the room and can tell the difference between a shout from the corner and a shout from the center.

The authors argue that if you ignore these differences, you might pick the wrong people to shout. You might ask someone far away to shout, which wastes their energy and doesn't help hide Alice, or you might ask someone too close, which creates too much noise for Bob.

The Solution: A Smart "On-Off" Switch

The paper proposes a clever new rule for choosing who in the crowd gets to shout. Instead of just picking people based on how well they can talk to Bob, the system looks at a special ratio: How loud is this person to Bob compared to how loud they are to Willie?

They call this the "activation metric."

If a person is quiet to Bob (so they don't mess up the conversation) but loud to Willie (so they confuse the guard), they get the green light to shout.
If they are loud to Bob or quiet to Willie, they stay silent.

This creates an "on-off" structure. The system sets a threshold, and anyone who fits the "perfect noise-maker" profile turns their signal ON at maximum power. Everyone else stays OFF. This is a huge improvement over older methods that tried to adjust everyone's volume slightly or picked people randomly.

What They Found: The Power of Scattering

Using complex math and computer simulations, the authors discovered some fascinating insights about how the crowd's layout affects the secret mission:

Scattering is Good: The more spread out the crowd is (spatial heterogeneity), the fewer people you need to shout to hide Alice. If the crowd is scattered, the guard sees a chaotic, unpredictable mix of noise. This chaos makes it very hard for him to tell if Alice's voice is hiding in there. The paper shows that a widely scattered group can reduce the number of required helpers significantly compared to a tightly packed group.
Distance Matters: If the crowd is, on average, very far from the guard, you need more people to shout. The further away they are, the weaker their noise is at the guard's ear, so you need a bigger crowd to create enough confusion.
The "Homogeneous" Trap: The paper proves that if you pretend everyone is the same distance from the guard (a "homogeneous" model), you will likely overestimate how many people you need. You might ask 100 people to shout when only 60 were actually necessary, which wastes energy and creates unnecessary noise for Bob.

The Math Magic: Finding the Sweet Spot

The researchers didn't just guess; they built a mathematical framework to find the exact number of people needed and the exact power Alice should use.

They derived a formula to calculate the minimum number of cooperative users needed to keep the guard confused.
They created a fast, low-complexity algorithm (a step-by-step recipe) to solve the power problem. Instead of trying millions of combinations (which would take forever), their method sorts the users and checks only a few specific "boundary" points. It's like finding the perfect temperature for a cake by checking the edges of the oven rather than testing every single degree.

The Results: Real-World Checks

The team ran simulations to see if their theory held up in a realistic, messy environment.

The Guard's View: They confirmed that as the number of observations (the length of the message) increases, the guard's ability to detect the signal approaches the theoretical limit.
Channel Errors: They also checked what happens if Bob doesn't know the exact quality of the connection to the crowd (imperfect channel estimation). They found that while this doesn't stop the guard from being confused, it does cause Bob to make mistakes in picking the right people. This leads to a slight drop in the speed (rate) of the secret message, but the secrecy itself remains intact.

Why It Matters

This paper shows that in the battle for wireless secrecy, location is everything. By treating the physical layout of the network as a design feature rather than just a random detail, we can make covert communication much more efficient. We can hide messages better using fewer helpers, saving energy and reducing interference. The authors suggest that future systems could use these ideas to protect sensitive data in military operations, secure IoT networks, and private communications, ensuring that even if someone is listening, they can't even tell that a conversation is happening.

Technical Summary: Covert Communication with Spatially Heterogeneous User Cooperation Against a Geometry-Aware Warden

Problem Statement
This paper addresses the challenge of covert communication in large-scale wireless networks where a single covert user (Alice) is assisted by multiple spatially distributed non-covert users against a "geometry-aware" warden (Willie). Unlike prior works that rely on receiver-centric activation, assume homogeneous channel conditions, or utilize averaged geometric parameters, this study focuses on a scenario where the warden is aware of the spatial distribution of users. The core difficulty lies in the fact that cooperative users causing similar interference at the legitimate receiver (Bob) may provide vastly different interference levels at Willie due to spatial heterogeneity. Existing designs that ignore these individual user-to-warden large-scale fading coefficients risk selecting ineffective users, leading to inaccurate estimates of the minimum detection error probability (DEP), unnecessary interference at Bob, and reduced achievable covert rates. The paper aims to develop a unified analytical framework that retains individual spatial characteristics to optimize cooperative user activation and power control.

Methodology
The authors propose a system model involving Alice, $M$ non-covert users, Bob, and Willie. The channel coefficients follow a Rayleigh distribution with user-dependent large-scale fading coefficients ( $\lambda_{m,b}$ and $\lambda_{m,w}$ ) determined by Euclidean distances. The study assumes a large-scale network ( $M \gg 1$ ) and utilizes a single-slope path-loss model.

The methodology proceeds through three main analytical stages:

On–Off Activation Structure: The authors derive an optimal activation rule where non-covert users are selected based on an activation metric $r_m = |h_{m,b}|^2 / \lambda_{m,w}$ . Users with $r_m$ below a threshold $\tau$ are activated to transmit interference at maximum power ( $P_{max}$ ), while others remain silent. This structure jointly balances the interference cost at Bob against the interference effectiveness at Willie.
Detection Analysis: Under the large- $M$ formulation, the paper characterizes the aggregate interference statistics at Willie. It derives a closed-form approximation for the DEP, the optimal detection threshold that minimizes this error, and the minimum number of cooperative users ( $K_{min}$ ) required to satisfy a covert constraint ( $\zeta_{min} \geq 1 - \epsilon$ ). The analysis accounts for the order-statistics-induced distribution of the selected user set.
Optimization Algorithm: The original high-dimensional power-control problem (optimizing $M+1$ variables) is reduced to a one-dimensional search over Alice's transmit power ( $P_a$ ). The authors develop a low-complexity piecewise search algorithm that evaluates only analytically relevant boundary candidates, avoiding the resolution dependence and computational burden of conventional grid searches.

Key Contributions
The paper outlines four primary contributions:

Spatial-Heterogeneity-Aware Design: The framework explicitly incorporates individual user-to-warden large-scale fading coefficients into both the cooperative user selection and the warden's detection analysis. This treats spatial heterogeneity as a design factor rather than a deployment detail.
Detection and Cooperation Analysis: The authors provide closed-form approximations for the DEP, the optimal detection threshold, and the minimum number of cooperative users. These expressions reflect how the activation rule alters the composition of the cooperative user set and its impact on aggregate interference uncertainty.
Low-Complexity Optimization: A piecewise search algorithm is proposed that reduces the optimization complexity to $O(M \log M)$ , dominated by a single sorting operation. This approach avoids the need for exhaustive grid searches and ensures the solution is not dependent on grid resolution.
Design Insights: The analysis reveals that greater spatial dispersion of users relative to Willie reduces the required number of cooperative users, whereas a larger average user-to-Willie distance increases this requirement. It also demonstrates that homogeneous or averaged-geometry analyses can overestimate cooperation requirements.

Results
Numerical results validate the theoretical analysis across various system parameters:

DEP Accuracy: The derived closed-form approximations for the DEP and optimal threshold closely match numerical simulations for both finite and large sample sizes.
Finite-Sample Behavior: The finite-sample DEP approaches the large-sample benchmark as the number of observations ( $N$ ) increases.
Impact of Spatial Parameters: Simulations confirm that increasing the standard deviation of user-to-Willie distances (spatial dispersion) decreases the minimum number of cooperative users needed. Conversely, increasing the mean distance increases the requirement.
Performance Gain: Compared to a baseline that selects users based solely on user-to-Bob channel gains (ignoring Willie's geometry), the proposed design achieves a higher covert rate and requires fewer cooperative users for the same covertness constraint.
Channel Estimation Errors: The study finds that imperfect user-to-Bob channel estimation does not significantly degrade the warden's detection performance (DEP remains similar to the perfect estimation case). However, it reduces the achievable rate at Bob due to activation errors, where users selected as weak interferers based on estimates may actually cause strong interference.

Significance and Claims
The paper claims that its primary significance lies in moving beyond the limitations of receiver-centric and homogeneous models. By retaining individual spatial characteristics, the proposed framework provides a more accurate characterization of the warden's detection capabilities and the necessary cooperation resources. The authors assert that their approach prevents the activation of unnecessary cooperative users, thereby minimizing interference at the legitimate receiver and maximizing the achievable covert rate. The work establishes that spatial heterogeneity, often treated as a nuisance, can be leveraged as a design factor to enhance covertness. The proposed low-complexity algorithm makes these analytical benefits practical for implementation in large-scale networks. The paper concludes modestly, noting that future work will extend the framework to mobile networks, multiple-warden scenarios, and integrated sensing and communication systems.