betaFeedback

WorksComputer Science

Low Probability of Target Detection in Wireless Sensing by Harnessing Reconfigurable-antenna based Reflectors

Abstract

In this paper, we introduce low probability of target detection (LPTD) as a new security objective for wireless sensing systems, where the existence of a target is concealed from passive wardens. Unlike conventional communication security and covert communication, LPTD aims to protect the sensed target itself from malicious detection, rather than concealing transmitted signals or communication activities. Specifically, we propose a reconfigurable-antenna-based reflector-assisted target covertness scheme, where passive reflectors deployed around the target are harnessed as artificial scattering sources to confuse the warden. By adjusting their reflection coefficients, these reflectors reshape the local scattering environment and mask the target-induced echo. We then analyze the warden's optimal detection strategy and derive the false-alarm and missed-detection probabilities, followed by a closed-form expression for its minimum total detection error probability. Numerical results validate the analysis and show that sufficient cooperation among reflectors can increase the warden's minimum total detection error probability to approximately 0.94–0.96, making target detection close to random guessing.

Introduction

Wireless sensing technologies have provided ubiquitous target awareness, but it also exposes physical targets to unauthorized observation. By reusing radio signals for detection, localization, and environmental perception, wireless sensing has become a promising technology for next-generation networks [1]. However, the target’s privacy in wireless sensing is inherently exposed to the open propagation environment. Since sensing waveforms and target echoes naturally radiate through space, a passive warden can detect the target’s presence and further estimate its sensitive sensing information by directly sensing the target’s physical response [2], leading to serious privacy leakage. Therefore, preventing the illegal detection of targets becomes a fundamental security challenge in wireless sensing systems.

Existing wireless security techniques cannot directly address this target detection threat, as they primarily focus on securing transmitted data or hiding communication behavior. Specifically, traditional physical-layer security methods, such as secure beamforming [3], artificial noise [4], and cooperative jamming [5], usually require the passive warden’s CSI or location for directional jamming design, and suffer from scalability issues due to extra power consumption or dedicated deployment [6]. Meanwhile, covert communications, also known as low probability of detection (LPD) communications, have been widely studied to hide confidential message transmission from a warden through techniques such as low-power transmission [7], spread-spectrum signaling [8], artificial noise, and backscatter-assisted reflection control [9]. However, since the warden detects the target by capturing its natural signal reflections, the target remains exposed regardless of how covert the transmitted message is. In summary, existing solutions are still insufficient to hinder the detection of targets in wireless sensing networks.

To protect the target from being detected, this work proposes a novel scheme that leverages reconfigurable-antenna-based reflectors to achieve low probability of target detection (LPTD) against wardens. Specifically, the proposed scheme harnesses widely distributed reflectors around the target, which can be implemented in various forms, such as ambient backscatter devices [10], reconfigurable surfaces [11], and reconfigurable corner reflectors [12], to act as artificial scattering sources to confuse the warden. By appropriately configuring their reflection coefficients or electromagnetic responses, these reflectors reshape the local scattering environment [13, 14], thereby reducing the statistical distinguishability between the target-induced scattering response and the surrounding reflected signals in the warden’s observation. In this way, the target echo is concealed within the uncertainty introduced by the surrounding reflections, making it difficult for a passive warden to reliably determine the presence of the true target. Specifically, the main contributions of this paper are summarized as follows.

  • We propose an reflector-assisted target covertness scheme to achieve LPTD against a target-detection warden. Specifically, we model a system where a multi-antenna monostatic radar transceiver illuminates the target area, while a passive warden (Willie) attempts to detect the target’s presence. The reflectors act as controllable reflectors to obfuscate Willie’s observation. Within this model, we extend the concept of LPD and formally define LPTD based on Willie’s minimum total detection error probability, which, to the best of our knowledge, is the first work in IoT sensing environments.

  • We formulate the target detection problem from the warden’s perspective and derive a closed-form expression for the minimum total detection error probability. By modeling a binary hypothesis-testing problem of target detection, we characterize Willie’s optimal target detection strategy, derive the false-alarm and missed-detection probabilities, and obtain a closed-form detection error boundary.

  • We verify the effectiveness of the proposed scheme through both theoretical derivation and simulation. The results demonstrate the effect of key system parameters on the LPTD performance, including the number of reflectors, transmit SNR, and scattering coherence. It is evident that sufficient reflector cooperation can increase the warden’s minimum total detection error probability to about 0.94–0.96, rendering Willie’s target detection nearly random guessing.

Notations: Boldface lower-case letters denote vectors. The operators (⋅)T(\cdot)^{T} and (⋅)H(\cdot)^{H} represent the transpose and the Hermitian transpose, respectively. ∣⋅∣|\cdot| denotes the absolute value of a complex scalar. E{⋅}\mathbb{E}\{\cdot\}, Pr⁡(⋅)\Pr(\cdot), and Pr⁡(⋅∣⋅)\Pr(\cdot\mid\cdot) represent the statistical expectation, probability, and conditional probability operators, respectively. x∼CN(μ,σ2)x\sim\mathcal{CN}(\mu,\sigma^{2}) indicates that the random variable xx follows a circularly symmetric complex Gaussian (CSCG) distribution with mean μ\mu and variance σ2\sigma^{2}.

Problem Formulation

System Model

As shown in Figure 1, we consider a wireless sensing system consisting of a legitimate base station (BS), one target, a target-detection warden named Willie, and KK reflectors deployed around the target. The BS transmits sensing waveforms to illuminate the target area, while Willie passively observes the wireless environment to determine whether the target is present. The reflectors are low-power passive devices that manipulate incident radio-frequency (RF) signals through controllable load impedances [15].

Figure 1. The system model, where a malicious target detection warden is confused by the signals from reflectors in sensing the existence of a true target.

The BS is equipped with MM transmit antennas, while Willie and each reflector have a single antenna [16]. The reflector set is denoted by K={1,…,K}\mathcal{K}=\{1,\ldots,K\}. A sensing frame contains LL symbols [17], where LL is sufficiently large such that noise fluctuations and signal-noise cross terms vanish. The BS transmit signal at the ℓ\ell-th symbol is

x[ℓ]=ws[ℓ],ℓ=1,…,L,(1)\mathbf{x}[\ell]=\mathbf{w}s[\ell],\quad\ell=1,\ldots,L, \tag*{(1)}

where w∈CM×1\mathbf{w}\in\mathbb{C}^{M\times1} is the beamforming vector satisfying ∥w∥2≤Pmax⁡\lVert\mathbf{w}\rVert^{2}\leq P_{\max}, and s[ℓ]s[\ell] is the normalized sensing symbol with 1L∑ℓ=1L∣s[ℓ]∣2=1\frac{1}{L}\sum_{\ell=1}^{L}|s[\ell]|^{2}=1.

All channels follow a block-fading model, remaining constant within each sensing frame and varying independently across frames. Let hBi∈CM×1\mathbf{h}_{Bi}\in\mathbb{C}^{M\times1} denote the BS-to-node channel for i∈{T,W,1,…,K}i\in\{T,W,1,\ldots,K\}, where TT, WW, and k∈Kk\in\mathcal{K} denote the target, Willie, and the kk-th reflector, respectively. Let hij∈Ch_{ij}\in\mathbb{C} denote the scalar channel between nodes ii and jj. Specifically, hBi=ℓ(dBi)h~Bi\mathbf{h}_{Bi}=\sqrt{\ell(d_{Bi})}\tilde{\mathbf{h}}_{Bi} and hij=ℓ(dij)h~ijh_{ij}=\sqrt{\ell(d_{ij})}\tilde{h}_{ij}, where h~Bi∼CN(0,IM)\tilde{\mathbf{h}}_{Bi}\sim\mathcal{CN}(\mathbf{0},\mathbf{I}_{M}) and h~ij∼CN(0,1)\tilde{h}_{ij}\sim\mathcal{CN}(0,1) are independent small-scale fading components. The large-scale path loss is modeled as ℓ(d)=C0(d/d0)−α\ell(d)=C_{0}(d/d_{0})^{-\alpha}, where C0C_{0} is the reference path loss at distance d0d_{0}, and α\alpha is the path-loss exponent.

Reflection Signal of Reflectors

The harnessed reconfigurable-antenna-based reflectors are deployed around the target to manipulate and reflect the incident RF signal. These reflectors can be implemented by ambient backscatter devices, RIS elements, or reconfigurable corner reflectors. Unlike active jammers, they do not generate new RF signals on their own. Instead, they reshape the surrounding scattering environment by controlling their electromagnetic reflection responses [18].

For a reconfigurable antenna, the reflected signal can be controlled by adjusting the load impedance connected to the antenna. Specifically, the antenna reflection coefficient is given by

Γk[ℓ]=RL,k[ℓ]−RA,k∗RL,k[ℓ]+RA,k,(2)\Gamma_{k}[\ell]=\frac{R_{L,k}[\ell]-R_{A,k}^{*}}{R_{L,k}[\ell]+R_{A,k}}, \tag*{(2)}

where RL,k[ℓ]R_{L,k}[\ell] denotes the configurable load impedance of the kk-th reflector at the ℓ\ell-th sensing symbol, RA,kR_{A,k} denotes its antenna impedance. By switching or continuously tuning RL,k[ℓ]R_{L,k}[\ell], the reflector can accordingly control the amplitude and phase of the reflected signal. Such impedance-based control provides a unified physical interpretation for different passive or nearly passive reflecting implementations.

Accordingly, we model the effective complex reflection coefficient of the kk-th reflector at the ℓ\ell-th sensing symbol as

βk[ℓ]=ρk[ℓ]ejϕk[ℓ],(3)\beta_k[\ell] = \rho_k[\ell]e^{j\phi_k[\ell]}, \tag*{(3)}

where ρk[ℓ]\rho_k[\ell] and ϕk[ℓ]\phi_k[\ell] denote the effective reflection amplitude and phase shift, respectively, 0≤ρk[ℓ]≤ρk,max⁡0 \leq\rho_k[\ell] \leq\rho_{k,\max}, ϕk[ℓ]∈[0,2π)\phi_k[\ell] \in[0,2\pi). The coefficient βk[ℓ]\beta_k[\ell] captures the overall configurable reflection response, including the effect of the antenna-load impedance and other implementation-dependent losses. Typically, the reflection coefficients remain fixed within a sensing frame, i.e., βk[ℓ]=βk\beta_k[\ell] = \beta_k, while they can be reconfigured across different frames to alter the local scattering environment.

The signal reflected by the kk-th reflector is therefore expressed as

xk[ℓ]=βk[ℓ]hBkHws[ℓ],(4)x_k[\ell] = \beta_k[\ell]\mathbf{h}_{B_k}^{H}\mathbf{w}s[\ell], \tag*{(4)}

where hBkHws[ℓ]\mathbf{h}_{B_k}^{H}\mathbf{w}s[\ell] represents the incident signal received by the kk-th reflector from the BS.

For notational convenience, the reflection coefficients of all KK reflectors at the ℓ\ell-th sensing symbol are collected into

β[ℓ]=[β1[ℓ],…,βK[ℓ]]T.(5)\boldsymbol{\beta}[\ell] = [\beta_1[\ell],\ldots,\beta_K[\ell]]^{T}. \tag*{(5)}

Willie's Target Detection

Willie’s goal is to infer the target’s existence from the received sensing signals. The general received signal at Willie during the ℓ\ell-th sensing symbol is given by

yW[ℓ]=hBWHws[ℓ]+αThTWhBTHws[ℓ]+∑k=1KhkWxk[ℓ]+nW[ℓ],(6)y_W[\ell] = \mathbf{h}_{BW}^{H}\mathbf{w}s[\ell] + \alpha_T\mathbf{h}_{TW}\mathbf{h}_{BT}^{H}\mathbf{w}s[\ell] + \sum_{k=1}^{K}h_{kW}x_k[\ell] + n_W[\ell], \tag*{(6)}

where αT∼CN(0,σT2)\alpha_T \sim\mathcal{CN}(0,\sigma_T^2) represents the radar cross section (RCS) [19] of the sensing target, and nW[ℓ]∼CN(0,σW2)n_W[\ell] \sim\mathcal{CN}(0,\sigma_W^2) is the additive white Gaussian noise (AWGN) at Willie. Due to the severe multiplicative path loss, the multi-cascaded signal between the target and the reflectors is neglected in the following performance analysis.

To detect the target, Willie formulates a binary hypothesis testing problem [20]. Define H0\mathcal{H}_0 as the target is absent, and H1\mathcal{H}_1 as the target is present. Therefore, Willie’s received signal under H0\mathcal{H}_0 and H1\mathcal{H}_1 is expressed as

yW[ℓ]={hBWHws[ℓ]+∑k=1KhkWβk[ℓ]hBkHws[ℓ]+nW[ℓ],H0,hBWHws[ℓ]+αThTWhBTHws[ℓ]+∑k=1KhkWβk[ℓ]hBkHws[ℓ]+nW[ℓ],H1.(7)y_W[\ell] = \begin{cases} \mathbf{h}_{BW}^{H}\mathbf{w}s[\ell] + \sum_{k=1}^{K}h_{kW}\beta_k[\ell]\mathbf{h}_{B_k}^{H}\mathbf{w}s[\ell] + n_W[\ell], & \mathcal{H}_0,\\ \mathbf{h}_{BW}^{H}\mathbf{w}s[\ell] + \alpha_T\mathbf{h}_{TW}\mathbf{h}_{BT}^{H}\mathbf{w}s[\ell] + \sum_{k=1}^{K}h_{kW}\beta_k[\ell]\mathbf{h}_{B_k}^{H}\mathbf{w}s[\ell] + n_W[\ell], & \mathcal{H}_1. \end{cases} \tag*{(7)}

Design Goals

The objective of our scheme design is to achieve LPTD at Willie, characterizing whether the reflectors can conceal the existence of the target from Willie. The reflectors reshape the scattering environment around the target, making the received energy at Willie under H0\mathcal{H}_0 and H1\mathcal{H}_1 difficult to distinguish. Specifically, we define LPTD as

Definition 1. Let PE∗P_E^* denote Willie’s optimal minimum total detection error probability after optimizing his detection threshold. The target is said to achieve LPTD (or covert target) if

PE∗≥1−ϵ,(8)P_E^* \geq1-\epsilon, \tag*{(8)}

where ϵ∈[0,1]\epsilon\in[0,1] specifies the desired LPTD or target covertness level.

Performance Analysis for Willie’s Detection Error Boundary

In this section, we derive Willie’s detection error boundary under the proposed reflector-assisted target covertness model.

Willie'sDetection Strategy

Since the BS sensing waveform is known, Willie can remove the deterministic direct BS-Willie component if the corresponding channel is available [21]. The residual received signal is defined as

y~W[ℓ]=yW[ℓ]−hBWHws[ℓ].(9)\widetilde{y}_W[\ell] = y_W[\ell] - \mathbf{h}_{BW}^{H}\mathbf{w}s[\ell]. \tag*{(9)}

Then, Willie performs a binary hypothesis test based on the residual observation vector y~W=[y~W[1],…,y~W[L]]\widetilde{\mathbf{y}}_W = [\widetilde{y}_W[1],\ldots,\widetilde{y}_W[L]]. According to the Neyman–Pearson lemma [22], for a given false alarm constraint, Willie’s optimal detector is the likelihood ratio test, given by

Λ(y~W)=fy~W∣H1(y~W∣H1)fy~W∣H0(y~W∣H0)≷H0H1η,(10)\Lambda(\widetilde{\mathbf{y}}_W) = \frac{f_{\widetilde{\mathbf{y}}_W|\mathcal{H}_1}(\widetilde{\mathbf{y}}_W|\mathcal{H}_1)}{f_{\widetilde{\mathbf{y}}_W|\mathcal{H}_0}(\widetilde{\mathbf{y}}_W|\mathcal{H}_0)} \mathop{\gtrless}_{\mathcal{H}_0}^{\mathcal{H}_1} \eta, \tag*{(10)}

where η\eta is the likelihood ratio threshold.

Since Willie does not know the instantaneous reflector reflection coefficients and cascaded reflector channels [13], he cannot coherently separate the target-induced echo from the reflector-induced reflections. Therefore, we consider a non-coherent radiometer detector, which is widely adopted in detection problems with unknown instantaneous CSI and a single antenna Willie [23]. Under this detector, Willie makes his decision based on the received residual energy, i.e., TW=1L∑ℓ=1L∣y~W[ℓ]∣2T_W = \frac{1}{L}\sum_{\ell=1}^{L}|\widetilde{y}_W[\ell]|^2. Therefore, the corresponding decision for the existence of the target is

TW≷H0H1τ,(11)T_W \mathop{\gtrless}_{\mathcal{H}_0}^{\mathcal{H}_1} \tau, \tag*{(11)}

where τ\tau is the energy detection threshold.

Let w=PBv\mathbf{w}=\sqrt{P_B}\mathbf{v}, where v\mathbf{v} denotes the transmit beamforming vector satisfying ∥v∥2=1\lVert\mathbf{v}\rVert^2=1. Define the equivalent reflector-induced scattering channel as gA=∑k=1KβkhkWHBkvg_A=\sum_{k=1}^{K}\beta_k\mathbf{h}_{kW}^{H}\mathbf{B}_k\mathbf{v}, and the direct target-induced scattering channel is defined as gT=αThTWHBTHvg_T=\alpha_T\mathbf{h}_{TW}^{H}\mathbf{B}_T^{H}\mathbf{v}, Willie’s residual signal can be written as

H0:y~W[ℓ]=PBgAs[ℓ]+nW[ℓ],H1:y~W[ℓ]=PB(gA+gT)s[ℓ]+nW[ℓ].(12)\begin{aligned} \mathcal{H}_0:\quad\widetilde{y}_W[\ell]&=\sqrt{P_B}g_As[\ell]+n_W[\ell],\\ \mathcal{H}_1:\quad\widetilde{y}_W[\ell]&=\sqrt{P_B}(g_A+g_T)s[\ell]+n_W[\ell]. \tag*{(12)} \end{aligned}

By the law of large numbers for the transmitted samples, the radiometer output converges to

H0:TW=PB∣gA∣2+σW2,(13)\mathcal{H}_0:\quad T_W=P_B|g_A|^2+\sigma_W^2, \tag*{(13)}
H1:TW=PB∣gA+gT∣2+σW2.(14)\mathcal{H}_1:\quad T_W=P_B|g_A+g_T|^2+\sigma_W^2. \tag*{(14)}

False Alarm and Missed Detection Probabilities

Although the channels themselves are independent, reflectors may exhibit spatial correlation or deliberate phase alignment, thereby resulting in a non-zero. To account for possible correlation among the reflector-reflected paths, we introduce a partial coherence factor χ∈[0,1]\chi\in[0,1] among different reflector-reflected paths. A larger χ\chi indicates stronger coherent combining among the reflector-reflected signals at Willie. Furthermore, we approximate the aggregate reflector-induced scattering channel as an equivalent circularly symmetric complex Gaussian random variable. This approximation is motivated by the central limit theorem [24], since gAg_A is formed by the superposition of multiple independent reflected paths with random amplitudes and phases. Thus, we have

gA∼CN(0,ΩA(K,χ,M)),(15)g_A\sim\mathcal{CN}(0,\Omega_A(K,\chi,M)), \tag*{(15)}

where ΩA(K,χ,M)\Omega_A(K,\chi,M) denotes the normalized average reflector-induced scattering strength. To capture the partial coherence among different reflector-reflected paths, define qk(M)=ρˉk2Gk(M)L(dBk)L(dkW)q_k(M)=\bar{\rho}_k^2G_k(M)L(d_{Bk})L(d_{kW}), where ρˉk2=E{∣βk∣2}\bar{\rho}_k^2=\mathbb{E}\{|\beta_k|^2\} is the average reflection power of the kk-th reflector, and Gk(M)G_k(M) is the effective BS antenna gain towards the kk-th reflector. Then, ΩA(K,χ,M)\Omega_A(K,\chi,M) is denoted as

ΩA(K,χ,M)=∑k=1Kqk(M)+2χ∑1≤i<j≤Kqi(M)qj(M).(16)\Omega_A(K,\chi,M)=\sum_{k=1}^{K}q_k(M)+2\chi\sum_{1\leq i<j\leq K}\sqrt{q_i(M)q_j(M)}. \tag*{(16)}

Similarly, we have

gT∼CN(0,ΩT(M)),(17)g_T\sim\mathcal{CN}(0,\Omega_T(M)), \tag*{(17)}

where ΩT(M)=σT2GT(M)L(dBT)L(dTW)\Omega_T(M)=\sigma_T^2G_T(M)L(d_{BT})L(d_{TW}), and GT(M)G_T(M) denotes the effective BS antenna gain towards the target.

Since gAg_A and gTg_T are independent zero-mean complex Gaussian random variables, their sum also follows a complex Gaussian distribution

gA+gT∼CN(0,ΩA(K,χ,M)+ΩT(M)).(18)g_A+g_T\sim\mathcal{CN}(0,\Omega_A(K,\chi,M)+\Omega_T(M)). \tag*{(18)}

Therefore, the random received powers under the two hypotheses follow exponential distributions

Z0=PB∣gA∣2∼Exp⁡(Zˉ0),(19)Z_0=P_B|g_A|^2\sim\operatorname{Exp}(\bar{Z}_0), \tag*{(19)}
Z1=PB∣gA+gT∣2∼Exp⁡(Zˉ1),(20)Z_1=P_B|g_A+g_T|^2\sim\operatorname{Exp}(\bar{Z}_1), \tag*{(20)}

where Zˉ0=PBΩA(K,χ,M)=σW2γBΩA(K,χ,M)\bar{Z}_0=P_B\Omega_A(K,\chi,M)=\sigma_W^2\gamma_B\Omega_A(K,\chi,M), Zˉ1=PB[ΩA(K,χ,M)+ΩT(M)]=σW2γB[ΩA(K,χ,M)+ΩT(M)]\bar{Z}_1=P_B[\Omega_A(K,\chi,M)+\Omega_T(M)]=\sigma_W^2\gamma_B[\Omega_A(K,\chi,M)+\Omega_T(M)], and γB=PBσW2\gamma_B=\frac{P_B}{\sigma_W^2} is the BS transmit SNR.

For a given threshold τ\tau, Willie’s false alarm probability is PFA,W(τ)=Pr⁡(TW>τ∣H0)P_{\mathrm{FA},W}(\tau)=\Pr(T_W>\tau\mid\mathcal{H}_0). Based on (14), we obtain

PFA,W(τ)=Pr⁡(Z0+σW2>τ).(21)P_{\mathrm{FA},W}(\tau)=\Pr(Z_0+\sigma_W^2>\tau). \tag*{(21)}

Since Z0Z_0 is exponentially distributed with mean Zˉ0\bar{Z}_0, the false alarm probability is

PFA,W(τ)={1,τ<σW2,exp⁡(−τ−σW2Zˉ0),τ≥σW2.(22)P_{\mathrm{FA},W}(\tau)= \begin{cases} 1, & \tau<\sigma_W^2,\\ \exp\left(-\frac{\tau-\sigma_W^2}{\bar{Z}_0}\right), & \tau\geq\sigma_W^2. \end{cases} \tag*{(22)}

Similarly, Willie’s missed detection probability is PMD,W(τ)=Pr⁡(TW≤τ∣H1)P_{\mathrm{MD},W}(\tau)=\Pr(T_W\leq\tau\mid\mathcal{H}_1). Based on (14), we have

PMD,W(τ)=Pr⁡(Z1+σW2≤τ).(23)P_{\mathrm{MD},W}(\tau)=\Pr(Z_1+\sigma_W^2\leq\tau). \tag*{(23)}

Since Z1Z_1 is exponentially distributed with mean Zˉ1\bar{Z}_1, the missed detection probability is

PMD,W(τ)={0,τ<σW2,1−exp⁡(−τ−σW2Zˉ1),τ≥σW2.(24)P_{\mathrm{MD},W}(\tau)= \begin{cases} 0, & \tau<\sigma_W^2,\\ 1-\exp\left(-\frac{\tau-\sigma_W^2}{\bar{Z}_1}\right), & \tau\geq\sigma_W^2. \end{cases} \tag*{(24)}

Total Detection Error Probability Boundary

The total detection error probability at Willie is

PE∗(τ)=PFA,W(τ)+PMD,W(τ).(25)P_E^*(\tau)=P_{\mathrm{FA},W}(\tau)+P_{\mathrm{MD},W}(\tau). \tag*{(25)}

Substituting (22) and (24) into (26), we obtain For τ≥σW2\tau\geq\sigma_W^2, define t=τ−σW2t=\tau-\sigma_W^2. Then, the total detection error probability is given by

PE(t)=1+exp⁡(−tZˉ0)−exp⁡(−tZˉ1),t≥0,(26)P_E(t)=1+\exp\left(-\frac{t}{\bar{Z}_0}\right)-\exp\left(-\frac{t}{\bar{Z}_1}\right),\quad t\geq0, \tag*{(26)}

and for τ<σW2\tau<\sigma_W^2, PE(τ)=1P_E(\tau)=1.

For τ≥σW2\tau\geq\sigma_W^2, define

t=τ−σW2.(27)t=\tau-\sigma_W^2. \tag*{(27)}

Then, Willie’s optimal probability of target detection can be obtained by setting the derivative of PE∗(t)P_E^*(t) to zero

∂PE∗(t)∂t=−1Zˉ0exp⁡(−tZˉ0)+1Zˉ1exp⁡(−tZˉ1)=0.\frac{\partial P_E^*(t)}{\partial t}=-\frac{1}{\bar{Z}_0}\exp\left(-\frac{t}{\bar{Z}_0}\right)+\frac{1}{\bar{Z}_1}\exp\left(-\frac{t}{\bar{Z}_1}\right)=0.

Then, we have

PE∗(t)=1+exp⁡(−tZˉ0)−exp⁡(−tZˉ1).(28)P_E^*(t)=1+\exp\left(-\frac{t}{\bar{Z}_0}\right)-\exp\left(-\frac{t}{\bar{Z}_1}\right). \tag*{(28)}

where t∗=Zˉ0Zˉ1Zˉ1−Zˉ0ln⁡(Zˉ1Zˉ0)t^{*}=\frac{\bar{Z}_{0}\bar{Z}_{1}}{\bar{Z}_{1}-\bar{Z}_{0}}\ln\left(\frac{\bar{Z}_{1}}{\bar{Z}_{0}}\right). According (27), Willie’s optimal energy threshold is

τ∗=σW2+Zˉ0Zˉ1Zˉ1−Zˉ0ln⁡(Zˉ1Zˉ0).(29)\tau^{*}=\sigma_{W}^{2}+\frac{\bar{Z}_{0}\bar{Z}_{1}}{\bar{Z}_{1}-\bar{Z}_{0}}\ln\left(\frac{\bar{Z}_{1}}{\bar{Z}_{0}}\right). \tag*{(29)}

Define the scattering separability ratio as

rW=Zˉ1Zˉ0=1+ΩT(M)ΩA(K,χ,M).(30)r_{W}=\frac{\bar{Z}_{1}}{\bar{Z}_{0}}=1+\frac{\Omega_{T}(M)}{\Omega_{A}(K,\chi,M)}. \tag*{(30)}

Since Zˉ1>Zˉ0\bar{Z}_{1}>\bar{Z}_{0}, we have rW>1r_{W}>1. Substituting t∗t^{*} into (28), Willie’s minimum total detection error probability is given by

PE∗=1+rW−rWrW−1−rW−1rW−1.(31)P_{E}^{*}=1+r_{W}^{-\frac{r_{W}}{r_{W}-1}}-r_{W}^{-\frac{1}{r_{W}-1}}. \tag*{(31)}

Performance Evaluation

In this section, we present numerical experimental results to evaluate the effectiveness of the proposed reflector-assisted LPTD scheme and to validate the derived detection error boundary. Based on the typical setting in wireless sensing and reflector [18], [25], [26], the BS, target, and Willie are located at (0, 0) m, (50, 0) m, and (40, 20) m, respectively. Passive reflectors are randomly distributed within a circular region of radius 5 m centered at the target. The BS is equipped with a 16-antenna ULA and steers its beam toward the target. The carrier frequency is set to 28 GHz, the path-loss exponent is α=2.2\alpha=2.2, and the reference path loss is given by C0=(λ4πd0)2C_{0}=\left(\frac{\lambda}{4\pi d_{0}}\right)^{2} with d0=1d_{0}=1 m. The noise floor at Willie is −90-90 dBm, and the BS transmit SNR is set to 10 dB. The average reflector reflection power and partial coherence factor are set to ρ2=0.1\rho^{2}=0.1 and χ=0.2\chi=0.2, respectively. We consider three target reflection strengths, i.e., σT2=1,3,5\sigma_{T}^{2}=1,3,5, corresponding to small, medium, and large targets. The number of reflectors varies from 1 to 29. Each result is averaged over 100 independent reflector layouts and 3×1043\times10^{4} Monte Carlo realizations.

Figure 2 compares the theoretical and simulation results of Willie’s minimum total detection error probability under different numbers of reflectors. The close agreement between the simulation markers and theoretical curves validates the derived analytical results. Moreover, PE∗P_{E}^{*}

Figure 2. Impact of reflector Number on LPTD Performance.

increases with KK because additional reflectors strengthen the surrounding scattering, reducing the statistical difference between H0\mathcal{H}_{0} and H1\mathcal{H}_{1} and thereby masking the target echo. Smaller targets achieve higher PE∗P_{E}^{*} since their weaker reflection produces a smaller ΩT\Omega_{T} and is therefore easier to conceal. Although larger targets are more detectable, increasing the number of reflectors still significantly improves the LPTD performance, demonstrating the effectiveness of the proposed target covertness scheme.

Figure 3 illustrates the impact of the transmit SNR γB\gamma_{B} on Willie’s detection performance. Without reflector assistance, a finite-block condition [27] is considered, where we find that increasing γB\gamma_{B} strengthens the target echo observed by Willie. As a result, the received energy distributions under H0\mathcal{H}_{0} and H1\mathcal{H}_{1} become increasingly separated, leading to a lower (PE∗)(P_{E}^{*}). This indicates that, in conventional wireless sensing without ambient scattering protection, a higher sensing transmit power directly benefits the passive target detector. In contrast, when reflectors are present, both the target-induced echo and the reflector-induced reflected signals are amplified by the BS transmit power. Hence, the dominant factor gradually changes from the absolute received power level to the relative scattering separability between the target echo and the reflector-induced ambient reflections. This explains why the reflector-assisted curves maintain much higher PE∗P_E^* over the considered SNR range.

Figure 3. Impact of Transmit SNR on LPTD Performance.

Figure 4 shows the effect of the coherence factor χ\chi among reflector-reflected paths. It can be seen that PE∗P_E^* increases as χ\chi becomes larger, especially when more reflectors are deployed. This is because a larger χ\chi enhances the coherent combination of the reflector-induced scattering components, thereby increasing the aggregate ambient scattering strength observed by Willie. As a result, the target-induced echo becomes less distinguishable from the surrounding reflected signals. The results indicate that not only the number of reflectors but also their scattering coherence plays an important role in determining the target covertness performance. In practice, this suggests that properly coordinating the reflection phases of reflectors can further enhance LPTD performance without increasing the BS transmit power.

Figure 4. Impact of reflector Scattering Coherence on LPTD Performance.

Overall, the above results consistently demonstrate the effectiveness of the proposed reflector-assisted LPTD scheme. By increasing the number of cooperative reflectors and enhancing the aggregate ambient scattering strength, the received energy distributions under the target-absent and target-present hypotheses become highly overlapped at Willie. Under the considered parameter settings, PE∗P_E^* can be increased to about 0.94–0.96, which means that Willie has almost no reliable capability to distinguish whether the true target exists. This result confirms that reflectors can effectively create a scattering-dominated uncertainty region around the target, thereby enabling a LPTD without active jamming or additional dedicated deployment.

Conclusion

This paper proposed a novel reflector-assisted target covertness scheme to achieve an LPTD at the warden of the malicious target detector. Rather than relying on active jamming, we harnessed distributed passive reflectors to reshape the local scattering environment, thereby overwhelming the target’s echo with controllable artificial uncertainty. Furthermore, we derived the closed-form boundary for the warden’s target detection error, revealing the fundamental relationship between the covertness and key system parameters. The results showed that Willie’s minimum total detection error probability can be increased to around 0.94–0.96 with sufficient cooperation among reflectors, making the warden’s target detection close to random guessing.

Acknowledge

This work was supported by the Smart Networks and Services Joint Undertaking under the EU’s Horizon Europe program (Ambient-6G, Grant No. 101192113), Business Finland (FISRE, Dnro 955/31/2026), and the Research Council of Finland (BANLOS, Decision No. 375519).

References

  1. [1]Y. Zhang, Y. Bai, S. Zeng, R. Jäntti, Z. Yan, C. Masouros, and Z. Han, “Backscatter device-aided integrated sensing and communication: A Pareto optimization framework,” IEEE Transactions on Wireless Communications, vol. 25, pp. 17958–17974, Jun. 2026.arxiv.org/abs/2507.09354
  2. [2]J. Zou, C. Masouros, F. Liu, and S. Sun, “Securing the sensing functionality in ISAC networks: An artificial noise design,” IEEE Transactions on Vehicular Technology, vol. 73, no. 11, pp. 17800–17805, Jul. 2024.arxiv.org/abs/2312.00981
  3. [3]Z. Zhang, Y. Liu, Z. Wang, X. Mu, and J. Chen, “Physical layer security in near-field communications,” IEEE Transactions on Vehicular Technology, vol. 73, no. 7, pp. 10 761–10 766, Feb. 2024.arxiv.org/abs/2302.04189
  4. [4]Y. Zhang, Z. Yan, J. Wang, Y. Yang, Z. Li, and R. Jäntti, “Artificial noise management for securing backscatter communication networks: Current advances, open challenges, and future directions,” IEEE Network, Early access, May 2025.DOI
  5. [5]X. Li, J. Jiang, H. Wang, C. Han, G. Chen, J. Du, C. Hu, and S. Mumtaz, “Physical layer security for wireless-powered ambient backscatter cooperative communication networks,” IEEE Transactions on Cognitive Communications and Networking, vol. 9, no. 4, pp. 927–939, Apr. 2023.DOI
  6. [6]J. Bai, H.-M. Wang, and L. Jin, “Dynamic agile reconfigurable intelligent surface antenna (DARISA) MIMO: DoF analysis and effective of optimization,” IEEE Transactions on Wireless Communications, vol. 25, pp. 2197–2212, Aug. 2025.arxiv.org/abs/2507.19136
  7. [7]Z. Yuchen, X. Sa, W. Jianquan, N. Boyu, Y. Xiaojun, and T. Wanbin, “Covert rate characterization under multi-antenna detection: A study on warden’s csi availability and noise uncertainty,” China Communications, vol. 22, no. 2, pp. 143–159, Apr. 2025.DOI
  8. [8]Z. Li, P. Hui, L. Guan, and C. Li, “A frequency hopping scheme for covert communication with noise uncertainty,” IEEE Transactions on Communications, vol. 74, pp. 3132–3146, Dec. 2025.DOI
  9. [9]Y. Zhang, R. Di Candia, H. Yigitler, R. Jäntti, and Z. Yan, “Covert backscatter communication in the presence of multi-antenna eavesdropper,” IEEE Communications Letters, vol. 28, no. 8, pp. 1770–1774, Jun. 2024.DOI
  10. [10]J. Famaey, O. L. A. López, R. A. Pérez, G. Callebaut, L. De Strycker, E. Katranares, L. Jiao, R. K. Singh, M. Ashraf, A. Azarbahram, A. Belogaev, J. Cappelle, A. Chassaigne, H. D. Moura, B. Deutschmann, M. C. Filippou, S. Jahanbazi, R. Jäntti, M. Kaveh, G. Leenders, U. Muehlmann, S. Nasser, P. P. Puluckul, D. Poehl, P. Raghuwanshi, L. Ramaut, O. M. Rosabal, D. E. Ruiz-Guirola, B. Singh, B. Xie, and Y. Zhang, “A survey on ambient iot: Use cases, standardization, enabling technologies, and research directions,” IEEE Communications Surveys & Tutorials, pp. 1–1, Aug. 2026.DOI
  11. [11]R. Deng, B. Di, H. Zhang, Y. Tan, and L. Song, “Reconfigurable holographic surface-enabled multi-user wireless communications: Amplitude-controlled holographic beamforming,” IEEE Transactions on Wireless Communications, vol. 21, no. 8, pp. 6003–6017, Jan. 2022.DOI
  12. [12]L. Gan, Z. Wu, and J. Li, “A passive isar jamming method based on time-modulated corner reflectors under asynchronous control,” IEEE Transactions on Aerospace and Electronic Systems, vol. 61, no. 5, pp. 11 103–11 115, Apr. 2025.DOI
  13. [13]Y. Zhang, Y. Yang, M. Kaveh, R. Jäntti, Z. Yan, D. Niyato, and Z. Han, “Ambshield: Enhancing physical layer security with ambient backscatter devices against eavesdroppers,” IEEE Transactions on Information Forensics and Security, vol. 21, pp. 4547–4562, Apr. 2026.
  14. [14]Y. Zhang, B. Xie, Y. Yang, Z. Yan, R. Jäntti, and Z. Han, “Authscatter: Accurate, robust, and scalable mutual authentication in physical layer for backscatter communications,” IEEE Transactions on Information Forensics and Security, vol. 20, pp. 6937–6952, Jul. 2025.DOI
  15. [15]Z. Y. Niya Luo, Yishan Yang and Y. Zhang, “Towards secure reflection: A survey on physical layer security in backscatter communications,” IEEE Communications Surveys & Tutorials, pp. 1–34, Jun. 2026.
  16. [16]C. Wang, Z. Li, X. Zhou, and L. Jin, “Intelligent cooperative covert исac via fluid antennas: A gan-optimized framework,” IEEE Journal of Selected Topics in Signal Processing, vol. 20, no. 3, pp. 262–276, Jan. 2026.DOI
  17. [17]Z. Zhang, K.-K. Wong, J. Dang, Z. Zhang, and C.-B. Chae, “On fundamental limits for fluid antenna-assisted integrated sensing and communications for unsourced random access,” IEEE Journal on Selected Areas in Communications, pp. 1–1, Sep. 2025.DOI
  18. [18]J. Famaey, O. L. López, R. A. Pérez, G. Callebaut, L. D. Strycker, E. Katranaras, J. Liao, R. K. Singh, M. Ashraf, A. Azarbahram et al., “A survey on ambient IoT: Use cases, standardization, enabling technologies, and research directions,” TechRxiv, Feb. 2026.DOI
  19. [19]Y. Zhang, Y. Bai, R. Jäntti, Z. Han, and C. Masouros, “Amb-sentry: Mitigating sensing eavesdropping in isac systems by harnessing ambient iot devices,” arXiv preprint arXiv:2608.11799, 2026.arxiv.org/abs/2608.11799
  20. [20]Y. Zhang, W. Ni, J. Wang, W. Tang, M. Jia, Y. C. Eldar, and D. Niyato, “Robust transceiver design for covert integrated sensing and communications with imperfect csi,” IEEE Transactions on Communications, vol. 73, no. 9, pp. 8016–8031, Apr. 2025.DOI
  21. [21]K. Han, K. Meng, and C. Masouros, “Sensing-secure ISAC: Ambiguity function engineering for impairing unauthorized sensing,” IEEE Transactions on Wireless Communications, pp. 1–1, Oct. 2025.DOI
  22. [22]H. V. Poor, An introduction to signal detection and estimation. Springer Science & Business Media, 2013.
  23. [23]T. V. Sobers, B. A. Bash, S. Guha, D. Towsley, and D. Goeckel, “Covert communication in the presence of an uninformed jammer,” IEEE Transactions on Wireless Communications, vol. 16, no. 9, pp. 6193–6206, Jun. 2017.arxiv.org/abs/1608.00698
  24. [24]D. Tse and P. Viswanath, Fundamentals of Wireless Communication. Cambridge, U.K.: Cambridge University Press, 2005.DOI
  25. [25]Y. Zhang, Y. Dang, M. Kaveh, Z. Yan, R. Jäntti, and Z. Han, “Physical layer challenge-response authentication between ambient backscatter devices,” IEEE Transactions on Information Forensics and Security, vol. 21, pp. 1512–1527, Jan. 2026.arxiv.org/abs/2506.18767
  26. [26]3rd Generation Partnership Project (3GPP), “Study on channel model for frequencies from 0.5 to 100 GHz,” 3GPP, Technical Report (TR) 38.901, 2020, release 16.
  27. [27]Z. Zhang, K.-K. Wong, J. Dang, Z. Zhang, C. Masouros, and C.-B. Chae, “On fundamental limits of slow-fluid antenna multiple access for unsourced random access,” IEEE Wireless Communications Letters, vol. 14, no. 11, pp. 3455–3459, Jul. 2025.DOI

Paper details

vlnc:W_9nanvu8m
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 7

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Computer Science

Not recorded for this paper

View statement
Looking for related papers…
Contents