Performance Analysis and Optimization of IRS-Aided Covert Communication With Hardware Impairments 阅读

Performance Analysis and Optimization of IRS-Aided Covert Communication With Hardware Impairments 阅读笔记

---

带有硬件损伤的信号模型

如图 1 所示，我们考虑一个借助 IRS 的典型隐蔽通信系统，其中 Alice 向 Bob 发送消息，而不会被看守威利检测到，看守威利试图检测是否发生了来自 Alice 的传输。

在这种情况下，我们假设 Alice、Bob 和 Willie 配备一根天线，而 IRS 配备 N 个无源反射元件。我们将 θ = diag(α1ejθ1 ,…,αNejθN ) 表示为 IRS 的对角反射矩阵，其中 0 ≤ αi ≤ 1 且 θi ∈ [0, 2π)，i = 1,…,N 表示反射幅度，分别为相移。

与现有的工作[6]、[7]、[8]、[9]、[10]不同，我们考虑HWI，它会导致收发器处的传输失真噪声和IRS处的相位误差。 IRS 处的 HWI 被建模为随机对角相位误差矩阵 [13]，可以表示为 θ~ = diag{ej~θ1 ,…,ej~θN }，其中 ~θn，对于 i = 1，。 …,N, 是随机相位误差均匀分布在 [−ψ,ψ] 上，其中 ψ ∈ [0,π) [14]。此外，由于加性失真噪声和幅度热噪声[13]，源和接收器处的HWI分别被建模为δt∼CN(0,Λ)和δr∼CN(0,Y)，其中Λ和Y是方差。从 Alice 到 Bob、IRS 和 Willie 的等效基带信道分别表示为 hab = √ χabhab、har = √ χarhar ∈ CN×1 和 haw = √ χawhaw。从 IRS 到 Bob 和 Willie 的通道分别表示为 hrb = √ χrbhrb ∈ CN×1 和 hrw = √ χrwhrw ∈ CN×1。另外，χij = β0( dij d0 )−αij ,ij ∈{ab, ar, aw, rb, rw} 表示大范围路径损耗，β0 为参考距离 d0 = 1m 处的信道功率增益，dij为节点 i 与节点 j 之间的距离，αij 为对应的路径损耗指数。 hab、haw、har、hrb、hrw 是相应的小尺度衰落系数，遵循零均值和单位方差的复杂高斯随机分布[15]、[16]。

Different from the existing works [6], [7], [8], [9], [10], we consider the HWIs, which cause transmitting distortion noise at transceivers and phase errors at IRS. The HWIs at IRS are modelled as a random diagonal phase error matrix [13], which can be expressed as $\tilde{\mathbf{\Theta }}=\mathrm{diag}\left\{e^{j{\tilde{\theta }}_1},\dots ,e^{j{\tilde{\theta }}_N}\right\}$ , where ${\tilde{\theta }}_n,fori=1,\dots ,N$ , are random phase errors uniformly distributed on $[-\phi ,\phi ]$ with $\phi \in [0,\pi )$[14] . Besides, due to the additive distortion noise and amplitude thermal noise [13], the HWIs at the source and receiver are modelled as ${\delta }_t\sim \mathcal{C}\mathcal{N}(0,\Lambda )$ and ${\delta }_r\sim \mathcal{C}\mathcal{N}(0,{\rm Y})$ , respectively, where $\Lambda$ and ${\rm Y}$ are variances.

The equivalent baseband channels from Alice to Bob, IRS and Willie are denoted by h_{a b}=\sqrt{\chi_{a b}} \bar{h}{a b}, \mathbf{h}{a r}=\sqrt{\chi_{a r}} \overline{\mathbf{h}}{a r} \in \mathbb{C}^{N \times 1} and h{a w}=\sqrt{\chi_{a w}} \bar{h}{a w} , respectively. And the channels from IRS to Bob and to Willie are denoted by \mathbf{h}{r b}=\sqrt{\chi_{r b}} \overline{\mathbf{h}}{r b} \in \mathbb{C}^{N \times 1} and \mathbf{h}{r w}=\sqrt{\chi_{r w}} \overline{\mathbf{h}}{r w} \in \mathbb{C}^{N \times 1} , respectively. In addition, \chi{i j}= \beta_{0}\left(\frac{d_{i j}}{d_{0}}\right)^{-\alpha_{i j}}, i j \in{a b, a r, a w, r b, r w} denotes the large-scale path loss, while \beta_{0} is the channel power gain at the reference distance d_{0}=1 m, d_{i j} is the distance between node i and node j , and \alpha_{i j} is the corresponding path loss exponent. \bar{h}{a b}, \bar{h}{a w}, \overline{\mathbf{h}}{a r}, \overline{\mathbf{h}}{r b}, \overline{\mathbf{h}}_{r w} are the corresponding small-scale fading coefficients, which follow the complex Gaussian random distribution with zero mean and unit variance [15], [16].

Suppose \mathcal{H}{0} denotes the null hypothesis in which Alice does not transmit, and \mathcal{H}{1} denotes the alternative hypothesis in which Alice transmits information to Bob. Therefore, the signal received at Bob and Willie can be written as