Physical layer key generation exploiting inherent channel randomness is an open research area in securing the networks with resource constraint nodes; therefore reduction of numerical computation is desirable to save battery power. However, the correlated components due to colored noise also affect the system performance. In this work, we consider the correlated colored noise components along with the additive white Gaussian noise (AWGN) in the wireless channel and analyze the effect of these correlated components on the system performance. We further propose a hybrid averaging and dimensionality reduction (AvDR), based received signal strength (RSS) preprocessing which is the combination of moving window averaging (Av) and principal component analysis (PCA) as dimensionality reduction technique (DR) to improve the system performance. Further, the system performance was evaluated by numerical simulations, and it is observed that the same improvement in system performance is achieved by generating keys from a fewer number of points selected after PCA as compared to processing all the points. Picking a few of the points in the data sequence instead of all reduces the total number of numerical calculations and saves system power, which is the primary requirement of resource constraint networks like the IoT.

中文翻译：

---

基于 AvDR 的无线安全密钥生成与物联网的有色噪声

利用固有信道随机性生成物理层密钥是一个开放的研究领域，用于保护具有资源约束节点的网络；因此需要减少数值计算以节省电池电量。然而，由于有色噪声引起的相关分量也会影响系统性能。在这项工作中，我们考虑了无线信道中的相关有色噪声分量以及加性高斯白噪声 (AWGN)，并分析了这些相关分量对系统性能的影响。我们进一步提出了一种基于接收信号强度（RSS）预处理的混合平均和降维（AvDR），它结合了移动窗口平均（Av）和主成分分析（PCA）作为降维技术（DR）来改进系统表现。进一步，通过数值模拟评估了系统性能，并且观察到，与处理所有点相比，通过在 PCA 之后选择的较少数量的点生成密钥来实现系统性能的相同改进。选择数据序列中的几个点而不是全部点可以减少数值计算的总数并节省系统功率，这是物联网等资源约束网络的首要要求。